DexCom, Inc. (NASDAQ: DXCM) — The Discount Closed Half the Gap; Now the Plan Has to Be Real

Date: August 8, 2026 Price reference: $84.75 (close, 2026-08-07) · Shares out ~377.3M · Market cap ~$32.0B · Net cash ~$0.55B · Enterprise value ~$31.4B Fiscal year: December · CIK 0001093557 · Sector: Health Care — Medical Devices (Diabetes / CGM) Coverage status: Update — follows on from my June 8, 2026 piece on the same name

⚡ Claude’s Take

This block is the author’s own subjective opinion. It is general information, not investment advice, and not a recommendation to buy or sell any security. The analysis that follows it is deliberately position-free and carries no price target — this block is the single exception.

Verdict: HOLD — downgraded from BUY/accumulate. A genuinely improving business whose easy money has been made; the long-range plan is now largely in the price. Conviction: medium. Tag: “I was right, and that is exactly why I am no longer buying it here.”

Two months ago I called DexCom a quality compounder de-rated on a self-inflicted stumble, with a margin trough visibly reversing, and set an accumulate zone of roughly $70–90 against a base case of $100–115. The thesis worked faster than I expected. The stock is $84.75, up 10.6% since the June 8 report and up 46.6% from its Investor Day close of $57.82 on May 14. Every load-bearing element of that call has now been confirmed by hard data: gross margin expanded a further ~400bp year-over-year in Q2 to 64.1% non-GAAP, trailing-twelve-month ROIC has doubled from 9.5% to 19.2% in five quarters, first-half free cash flow more than doubled to $633.5M, the company put $600M of buyback to work at an average of ~$69.77 — its largest quarterly repurchase ever, and pointedly counter-cyclical — and the CONNECT trial delivered the strongest possible rebuttal to the GLP-1 bear case, with the GLP-1-only cohort showing the largest A1C benefit from CGM.

So why downgrade? Because valuation is a moving target and this one moved. In June the stock sat in the bottom ~3rd percentile of its own ten-year valuation range; it now sits in roughly the bottom 10th (composite 9.8th percentile, P/B 9.6th, P/S 11.6th). More importantly, the arithmetic has changed character. At an EV of ~$31.4B against $1.14B of trailing EBIT — 27.6x — the market is now underwriting close to the full delivery of management’s own 2030 long-range plan: 10%+ organic growth every year and a 29–30% operating margin. In May that plan was free optionality; today it is the base case. That is a materially worse bargain even though the business is materially better. My revised view: hold what you own; the accumulate zone drops to roughly $65–75 (a re-test, not a forecast), and I would want the CMS non-insulin decision in hand before paying above the high-$90s. Base-case fair value is unchanged at ~$115–130 on the LRP being delivered; the bear case is no longer a large loss but a flat five years — a $7.0B/25%-margin 2030 gets you back to roughly today’s price. This has stopped being a contrarian-value setup and become a quality-compounder-at-a-fair-price, in the early stage of momentum, with a binary regulatory catalyst pending.

What flips me bullish again: a favourable CMS national coverage decision for non-insulin Type 2 before year-end (management expects a decision this year, effective mid-2027) — that is a genuine step-change in TAM that the current multiple does not fully capitalise, and it would put the bull case of ~$165 in play. What flips me bearish: any evidence the gross-margin recovery has run its course — the Ireland factory start-up already caps second-half expansion, and if 2027 guidance shows margin flattening below the LRP glidepath, the entire 27.6x EBIT multiple loses its justification. Secondary bear trigger: escalation of the still-unresolved March-2025 FDA warning letter, which the Q2 10-Q is conspicuously silent about. Not a short — the business is compounding and the buyback is real. Just no longer a bargain.

📈 Stock Price Action — Five-Year Event Map

DexCom’s five years are a full round trip and a partial recovery. The stock peaked at $164.86 intraday in November 2021, fell through a two-and-a-half-year de-rating punctuated by a single catastrophic session, bottomed at $54.11 in November 2025, and has since retraced to $84.75 (close, 2026-08-07) — still 48.6% below the five-year high but 56.6% above the low. The 52-week range is $54.11–$87.83, with the high set on 2026-08-04, four days before this report. Year-to-date the stock is +27.7% (from $66.37 at 2025 year-end), making it the best-performing large-cap medical-device name among its peers this year.

Cycle narrative. (1) The 2021–22 decline was a multiple event, not a business event — Dexcom grew revenue throughout while a 30x-sales biotech-adjacent multiple compressed with rates. (2) 2023 brought a genuine fundamental catalyst in Medicare’s April-2023 coverage expansion, but the stock ended 2023 roughly where it started 2023. (3) July 26, 2024 is the defining event of the period: a −40.7% single session, the largest in company history, after management cut full-year guidance following a sales-force reorganisation that collided with the DME-to-pharmacy channel transition. That was a self-inflicted execution failure, and it is why the market subsequently refused to pay a premium multiple. (4) The 2024–25 grind lower layered three separate problems onto the credibility loss: gross margin fell to a 57.5% GAAP trough in Q1 2025, the FDA issued a warning letter in March 2025 citing deficient responses to prior Form 483 observations, and the GLP-1 “diabetes-is-cured” narrative took hold. (5) Two further breaks in autumn 2025 — the September 19 −11% session and the October 31 −14.6% post-print session — carried the stock to its $54.11 low on November 7. (6) The 2026 recovery began at the Investor Day, not at an earnings print. From $57.82 on May 14 the stock added 23.6% in four sessions as management laid out the first credible multi-year financial framework of the post-Sayer era. (7) The CONNECT readout on June 6 and the first-ever pediatric OTC CGM clearance on June 12 added clinical substance. (8) The Q2 beat-and-raise on July 30 (+12.0% the following session) confirmed rather than initiated the move, carrying the stock to a 52-week high of $87.83 on August 4. Price moves above are facts drawn from the AZI adjusted-price series; the attributed drivers are interpretations cross-referenced against the corresponding 8-Ks, earnings releases and news items.

0.5 Changes Since June 8, 2026

Nine things changed materially in the two months since the earlier report. Three of them change the thesis.

1. Q2 2026 was a decisive beat-and-raise (thesis-confirming). Revenue of $1,308.4M grew 13.1% reported and 12% organic. Non-GAAP gross margin reached 64.1%, up ~400bp year-over-year; non-GAAP operating margin reached 25.1% versus 19.2%. Non-GAAP EPS of $0.70 beat the $0.61 consensus by 14.8% and grew 46%. Management raised FY2026 guidance on every line: revenue to $5.18–5.25B (11–13% growth), non-GAAP gross margin to approximately 64%, operating margin to 23.5–24% (from 23–23.5%), adjusted EBITDA margin to 31.5–32%.

2. The Investor Day long-range plan — which the earlier report did not cover (thesis-changing). On May 14, 2026 Dexcom held its Investor Day and, for the first time, published a dated five-year financial framework: organic revenue growth above 10% in every year through 2030 (explicitly not an average CAGR), a 2030 non-GAAP gross margin in the high-60s (67–69%), a 2030 non-GAAP operating margin of 29–30%, and a 2030 adjusted EBITDA margin of 36–37% — with CMS competitive bidding assumed to begin in 2028 and already embedded in those numbers. The earlier report, written 25 days later, contains no reference to any of this and left it as an open question. It should not have; this is the most important disclosure of the year and it is what re-rated the stock.

3. Capital allocation was materially upgraded (thesis-changing). The board terminated the 2025 $750M program with $250M unused and authorised a $1.00 billion 2026 program, of which $600.0M was executed in Q2 at an average of ~$69.77 per share — 8.6 million shares, the largest quarterly repurchase in company history, and bought roughly 21% below the current price. The 10-Q additionally establishes a standing policy: “at least 50% of cash generated from operations, net of planned capital expenditures, to share repurchases over the course of our long-range plan.” The earlier report’s central capital-allocation criticism — pro-cyclical buyback timing — is now falsified by the record.

4. CONNECT delivered, and it hit the GLP-1 bear case directly (thesis-confirming). The full randomised readout at ADA on June 6 showed a 1.6% A1C improvement in the G7 arm, a 0.9% difference versus routine care, more than five additional hours per day in range, and 97% median wear over 26 weeks. Critically, the largest relative A1C improvement occurred in the cohort using GLP-1 therapy alone. Dexcom has submitted the data to CMS.

5. A CMS decision is now dated (new catalyst). Management expects a CMS decision on non-insulin Type 2 coverage before the end of 2026, with coverage assumed effective mid-2027 in the plan. The addressable population is ~25 million in the US, roughly half of Medicare age.

6. Commercial coverage is effectively complete in the private market. As of summer 2026 all four largest US commercial PBMs cover Dexcom CGM for all people with diabetes — more than 7 million Type 2 non-insulin lives. Management also disclosed that ~9 million Americans already have CGM coverage and are not using it, which relocates the near-term constraint from payers to prescriber behaviour.

7. First acquisition of any note. Dexcom acquired Nutrisense in June 2026 for $41.2M net of cash — a CGM-guided nutrition-insight platform and a long-standing partner. Most of Nutrisense’s revenue was already Dexcom Stelo pass-through; incremental non-CGM revenue is “the millions, you can count on one hand.” Immaterial financially, but it marks the start of a stated tuck-in M&A appetite.

8. Product and regulatory milestones stacked up. Health Canada cleared G7 15-Day on July 13 (first ex-US regulator); Dexcom Flex launched in Germany; Stelo won the first-ever OTC CGM pediatric clearance on June 12 and a fully rebuilt app in late July; and Dexcom was named on July 22 the first company selected for the FDA’s TEMPO digital-health pilot. G8 remains on track for 2027 submission and end-2027/early-2028 launch.

9. The falsification tests from the earlier report — scorecard. The earlier report’s bull test was “two more quarters of gross margin holding 63%+ and non-insulin/Stelo volume acceleration.” Q1 2026 GAAP gross margin was 62.95%, Q2 was 63.44% — the test is passed on margin (63%+ on the non-GAAP basis management guides to, 64.1% in Q2) and partially passed on volume (record new-patient starts sustained, but Stelo unit disclosure remains absent). The bear test was FDA escalation toward a consent decree or an Abbott price war; neither has occurred, and the Q2 10-Q does not mention the warning letter at all.

Net effect on the thesis. The business improved on every axis the earlier report identified as load-bearing, and management supplied the credible multi-year framework whose absence was the earlier report’s biggest open question. What has deteriorated is only the price. The correct response is to keep the fundamental verdicts and mark down the valuation attractiveness — which is what this update does.

1. Executive Summary

DexCom is the value leader in continuous glucose monitoring, and after two years of self-inflicted damage the franchise is visibly repairing. In the twelve months to June 30, 2026 it generated $4,969M of revenue at a 62.5% gross margin and a 22.9% operating margin, produced $999.7M of GAAP net income, and earned a 19.2% return on invested capital and a 38.5% return on equity. In the first half of 2026 alone free cash flow reached $633.5M against $305.7M a year earlier, a 107% increase on 12% revenue growth. The balance sheet holds $1,947M of cash and short-term securities against $1,398M of debt — net cash of roughly $549M.

The recovery is not a rebound narrative; it is visible in six consecutive quarters of expanding trailing-twelve-month margins. Trailing gross margin has run 58.8% → 59.0% → 60.1% → 61.5% → 62.5%; trailing operating margin 16.0% → 17.2% → 19.6% → 21.5% → 22.9%; trailing ROIC 9.8% → 11.8% → 15.8% → 17.2% → 19.2%. Trailing incremental operating margin is 67.6%, meaning roughly two-thirds of each incremental revenue dollar over the past year converted to operating profit. Management attributes the improvement to reduced scrap, manufacturing throughput and the mix shift to the 15-day sensor — process gains, not price.

Two disclosures in the period reset the framework. At the May 14, 2026 Investor Day management published its first dated long-range plan: organic growth above 10% every year through 2030, a 2030 non-GAAP gross margin in the high-60s, a 29–30% operating margin, and a 36–37% adjusted EBITDA margin, with CMS competitive bidding assumed from 2028. And it committed capital to it — a $1.0B repurchase authorisation with $600M already executed in Q2 at ~$69.77 per share, plus a standing policy of returning at least half of free cash flow to buybacks. On the clinical side, the CONNECT randomised trial delivered a 0.9% A1C advantage over routine care in non-insulin Type 2 patients with 97% median wear, and — directly contradicting the GLP-1 bear case — the largest benefit appeared in the GLP-1-only cohort.

The stock has responded. It is up 27.7% year-to-date and 46.6% from the Investor Day close, made a 52-week high of $87.83 on August 4, and has moved from the bottom ~3rd percentile of its own ten-year valuation range to roughly the bottom 10th. At $84.75 the enterprise is capitalised at ~$31.4B, or 27.6x trailing EBIT, 22.4x trailing EBITDA, 6.3x trailing revenue, 33.0x FY2026 consensus EPS and 27.8x FY2027. On my arithmetic, that price already embeds close to full delivery of the 2030 plan.

The risks are unchanged in character but changed in weight. The moat — customer captivity via the automated-insulin-delivery ecosystem plus scale economics in sensor manufacturing — is real, and the scale leg is visibly widening; but pump partners still support Abbott alongside Dexcom, and the profit pool continues to migrate toward lower-ASP, lower-defensibility Type 2 and OTC tiers. CMS competitive bidding is now a scheduled 2028 price-down, not a hypothetical. The March-2025 FDA warning letter remains unmentioned in the Q2 10-Q and unresolved on the public record, and the company faces a securities class action covering January 8, 2024 to September 17, 2025, derivative actions, and G6/G7 product class actions, with no loss accrual disclosed. Insider activity over the period was uniformly 10b5-1 selling with no open-market purchases.

Net: a good business getting demonstrably better, run by a management team that has now published a plan and backed it with the largest buyback in its history — trading at a price that assumes the plan works. No recommendation or price target appears below this section.

2. Business Overview

What DexCom does. Dexcom designs, manufactures and sells continuous glucose monitoring systems: small wearable sensors that measure interstitial glucose every few minutes and stream readings to a smartphone, smartwatch, receiver or insulin pump, displacing the episodic fingerstick. The disposable sensor is worn 10 to 15.5 days; the economics are the recurring reorder of that consumable. With G7 the transmitter is integrated into the disposable, so essentially the entire device is a recurring purchase. (FACT — FY2025 10-K; product disclosures.)

Product portfolio as of Q2 2026.

How it makes money. Overwhelmingly recurring sensor consumables sold through pharmacy and DME distributors, with a smaller direct channel. Revenue recognition follows shipment to the distributor; the underlying economics are a patient installed base reordering monthly. This is a razor/razor-blade structure in which the blade is essentially the whole razor.

Revenue segmentation (Q2 2026). By geography, US revenue was $933M, up 11%, and international revenue was $375M, up 19% reported and 16% organic — international is now 28.7% of the total and is growing at roughly 1.5x the US rate. Management specifically cited France and Canada, markets where access recently expanded, as the strongest performers, which is the operative pattern internationally: reimbursement wins, then tender wins, then share.

Customer tiers, in descending order of economic quality. (1) Intensive insulin users — Type 1 and insulin-using Type 2 — the high-ASP, high-retention core, anchored by AID pump integration. (2) Basal and non-insulin Type 2 — the reimbursed growth frontier; ~7 million lives now covered across the four largest commercial PBMs, with a ~25 million US population pending a CMS decision. (3) Prediabetes and wellness — the cash-pay OTC frontier addressed by Stelo, against a ~115 million US prediabetes population of whom only a fraction are diagnosed. Defensibility, ASP and gross margin all decline as you move down the tiers; volume potential rises.

Recurring versus non-recurring. Predominantly recurring reorder revenue with high visibility. Global new-customer starts in Q2 were in line with Q1’s record, with a sequential increase in the US and a modest decline outside it — management attributed the international softness to tender timing. Notably, the US strength was broad-based: management reported step-ups in Type 1, intensive Type 2, basal, and Type 2 non-insulin simultaneously, rather than a single segment carrying the quarter.

Verdict (Business Overview): A focused, recurring-revenue medical-device franchise with a clean consumable model, a long secular substitution tailwind, and — new this period — a materially broadened product ladder that now reaches from AID-integrated intensive insulin users down to over-the-counter pediatric wellness. The portfolio breadth is a genuine improvement on the prior year: Flex, Stelo pediatric, and the 15-day platform give Dexcom a price-tiered answer to Abbott in every segment rather than a premium-only position. The unchanged nuance is that incremental growth still migrates toward lower-ASP, lower-defensibility tiers.

3. Industry Dynamics

Market size and growth. Third-party estimates put the global CGM market at roughly $11–13B in 2025 growing ~15% annually into the early 2030s, with sensors — the recurring consumable — approximately 84% of value and North America roughly half. The structural driver is the multi-decade substitution of continuous monitoring for episodic fingerstick testing, now compounded by reimbursement expansion into non-insulin populations and the emergence of over-the-counter use. (FACT — industry sources as cited in the earlier report and re-verified in the source appendix.)

The population runway is not the constraint; awareness and reimbursement are. ~38 million Americans and ~590 million adults globally have diabetes, projected toward ~640 million by 2030, atop a far larger prediabetes pool of roughly 115 million in the US alone. CGM penetration is high in Type 1, still climbing in insulin-using Type 2, low in non-insulin Type 2 and negligible in prediabetes. The most instructive new datum this quarter is management’s disclosure that approximately 9 million people in the US already have CGM coverage and are not using it. (FACT — Q2 2026 earnings call.)

Interpretation — this reframes the near-term growth debate. For two years the market has argued about whether payers would extend coverage. That argument is largely settled in the commercial market: all four largest PBMs now cover all people with diabetes. The binding constraint has moved to physician awareness and prescribing workflow — which is an execution problem inside Dexcom’s control (sales-force claims-adjudication tools, prior-authorisation removal, CONNECT-based detailing) rather than a payer problem outside it. That is a strictly better problem to have, and it is why the US grew 11% with a sequential uptick in new starts despite no new coverage event in the quarter.

Profit pools versus volume pools — the persistent structural tension. The most penetrated segments (Type 1, intensive Type 2) carry the highest ASP, the highest margin and the strongest switching costs. The growth segments (basal Type 2, non-insulin Type 2, OTC/prediabetes) carry lower ASP, thinner or no reimbursement and weaker switching costs. Volume TAM continues to expand faster than profit-pool TAM. The mitigating development is that Dexcom’s cost curve is falling fast enough to defend margin through the mix shift: the 15-day sensor lowers cost-per-day-of-wear structurally, and G8 is explicitly designed to be cheaper to manufacture than G7. Gross margin rising 400bp year-over-year while mix shifted toward lower-ASP tiers is the empirical proof that the cost curve is currently winning. (INTERPRETATION.)

Competitive intensity — a duopoly, still. CGM remains effectively a Dexcom/Abbott duopoly, with Medtronic, Senseonics/Ascensia and Roche minor. Dexcom leads US value share by premium positioning; Abbott leads global unit volume through FreeStyle Libre’s lower-price, mass-market strategy. Nothing in the period suggests a structural escalation: Dexcom took share in Q2 across multiple patient categories on its own account, and — significantly — did so while expanding gross margin, which is not the signature of a price war. The counter-observation is that Abbott’s twelve-month relative strength is −15.8%, so the market is not treating Abbott as the winner either. (FACT/INTERPRETATION.)

Regulation and reimbursement — the vector now cuts both ways on a schedule. Three regulatory developments in the period matter.

• CMS non-insulin Type 2 coverage. Dexcom has submitted CONNECT to CMS. Management expects a decision before the end of 2026 and has modelled coverage effective mid-2027. This is the single largest discrete swing factor in the story, addressing ~25 million Americans of whom roughly half are Medicare-age.
• CMS competitive bidding. The June-2025 proposal to subject CGMs to DMEPOS competitive bidding is no longer an open-ended threat: management has stated the long-range plan assumes competitive bidding begins in 2028 and that the 10%+ organic growth target holds through it. That is a meaningful disclosure — it converts an unquantified risk into a dated, budgeted headwind.
• FDA TEMPO. Dexcom was named on July 22, 2026 the first participant in the FDA’s TEMPO digital-health pilot, a framework granting a form of enforcement discretion for digital health technologies in early-stage cardiometabolic conditions. Management was careful to say TEMPO is not a coverage mechanism and is unconnected to the broad Type 2 CMS decision — it accelerates innovation cycles, particularly around prediabetes screening. I take management at their word on the limitation and treat TEMPO as a modest regulatory-relationship positive, not a revenue event.

Capital-cycle read (Marathon lens). The earlier report placed CGM in a mid-to-late capital cycle with returns mean-reverting, and read the 2023→2025 gross-margin compression as that signature made visible. Six quarters of margin expansion and a doubling of ROIC argue that reading was premature — the compression was substantially company-specific (scrap, freight, quality remediation) rather than industry-wide price competition. The revised read: returns are inflecting upward now, driven by idiosyncratic operating improvement, with a known industry-wide price-down scheduled for 2028. Capital continues to be attracted — Medtronic sourcing an Abbott-built sensor, Senseonics pushing implantables, OTC entrants, and the perennial Apple/Samsung non-invasive optical threat — but none of it has yet produced the ASP collapse the capital-cycle framework would predict. The framework’s warning remains live for 2028 onward.

Verdict (Industry Dynamics): Structurally GOOD, and better than the earlier report allowed. The industry is large, secular, consumable-driven and growing mid-teens; reimbursement in the commercial market is now essentially won; the near-term constraint has shifted to a controllable execution variable; and the principal structural price-down has a date attached and is already embedded in the incumbent’s plan. The unchanged negative is that the profit pool per patient shrinks as the mix moves down-tier, and that a two-player market with a scheduled government price-setting mechanism in 2028 is not a market where anyone will earn a monopoly return forever. I upgrade the industry read from “good but deteriorating at the margin” to “good, with the deterioration deferred and quantified.”

4. Competitive Position

Name the moat. Dexcom’s advantage is a combination of two of Greenwald’s three genuine types: customer captivity and economies of scale. There is no brand or network effect worth the name.

Customer captivity operates through the automated-insulin-delivery ecosystem. An intensive insulin user on a Tandem, Insulet or Beta Bionics pump running a closed-loop algorithm calibrated to a Dexcom sensor faces real switching friction: retraining, re-pairing, clinical revalidation, and the loss of a data history their endocrinologist reads. Empirically, this shows up as high wear-time and reorder persistence — in CONNECT, median wear was 97% over 26 weeks, and management reports above-80% utilisation in its real-world commercial non-insulin registry. That is captivity you can measure.

Economies of scale operate in sensor manufacturing, and this is the leg that visibly strengthened in the period. Gross margin expanded ~400bp year-over-year to 64.1% non-GAAP, and the CFO attributed the improvement specifically to reduced scrap, manufacturing throughput and quality-management gains, plus the initial 15-day mix benefit. Scrap reduction is process learning on high-volume automated lines — a genuine scale advantage, because the learning rate is a function of cumulative units produced, and Dexcom and Abbott are the only two firms producing at that cumulative volume. G8’s explicit design goal of being cheaper to manufacture than G7 extends the same logic.

The moat test that matters: would a financial outcome deteriorate without it? Yes, and I can name it. Without AID captivity, Dexcom’s premium ASP in the intensive-insulin tier would converge toward Abbott’s mass-market price point, compressing a gross margin that is currently 64% toward the low-50s. Without manufacturing scale, the 15-day transition would be margin-neutral rather than margin-accretive, because the cost saving comes from producing fewer, longer-lived sensors on lines whose fixed cost is already absorbed. Both are testable and both currently hold.

Where the moat is narrow — unchanged from the earlier report. Pump partners are not exclusive. Tandem, Insulet and Beta Bionics all support Abbott sensors alongside Dexcom. Medtronic has gone further and sources a sensor built by Abbott for the 780G. The AID ecosystem is therefore a switching cost, not a lock-out: it makes leaving Dexcom annoying, not impossible, and every additional Abbott integration lowers the annoyance. Nothing in the period improved this.

Head-to-head versus Abbott. Abbott has matched Dexcom on accuracy and beaten it on price and unit volume; FreeStyle Libre passed 6 million users. Dexcom’s counter is accuracy leadership at the top of the range (G7 15 Day at MARD 8.0%, marketed as the longest-lasting and most accurate CGM cleared by Health Canada), AID integration depth, and now a full price ladder — Flex and ONE+ for tender-driven international value segments, Stelo for OTC. The strategic change this year is that Dexcom stopped trying to defend a premium-only position and built a tiered portfolio. That is the right answer to a volume competitor, and it is what makes the international 16% organic growth possible. It is also, mechanically, ASP-dilutive, which is why the cost curve has to keep falling.

Head-to-head versus the rest. Senseonics’ implantable Eversense is a niche product with a clinician-implantation friction that caps its addressable share. Medtronic’s decision to source Abbott’s sensor is an admission it cannot build a competitive one — good for the duopoly structure, bad in that it puts Abbott silicon inside Medtronic’s installed base. The Apple/Samsung non-invasive optical threat remains perennially five years away and has produced no cleared product; I treat it as a tail risk with a long fuse, not a base-case competitive factor.

A market-share-stability test (Greenwald). Stable share over long periods is the strongest evidence of a moat. Here the evidence is mixed, and worth stating honestly: Dexcom lost share in 2024–25 through its own commercial failure and is now taking it back — management reported share gains across several core markets and patient categories in Q2. Share that can be lost through a sales-force reorganisation and regained through better customer service is not the signature of an impregnable moat; it is the signature of a strong but contestable position in a two-player market. (INTERPRETATION.)

Verdict (Competitive Position): A real but contestable moat, with the cost leg widening. Dexcom has genuine customer captivity in its most valuable tier and a genuine, currently widening, manufacturing scale advantage. It does not have a lock-out, it does not have exclusivity with its AID partners, and it operates against a competitor with more units and a lower price. The honest characterisation is a strong number-two-by-volume, number-one-by-value position whose defensibility is highest exactly where the growth is slowest. I hold the earlier report’s “real but narrow” verdict, with one upgrade: the evidence that the scale/cost advantage is compounding — 400bp of margin in a year, driven by process rather than price — is stronger than it was in June.

5. Growth History and Forward Opportunities

Historical growth. Revenue: FY2024 $4,033M, FY2025 $4,662M (+15.6%), and TTM through Q2 2026 $4,969M. The quarterly sequence tells the more useful story:

Two readings. First, growth has settled into a low-teens band — 13.1% in Q2 2026, 12% organic — which is a deceleration from the ~26% nine-year CAGR the earlier report cited and is the honest run-rate to underwrite. Second, and more importantly, margin expansion has decoupled from growth: Q2 2026 grew slower than Q3 2025 but earned 290bp more operating margin. That is operating leverage, and it is why earnings growth (+46% non-GAAP EPS) is running at roughly 3.5x revenue growth.

Organic versus acquired. Growth is essentially entirely organic. The Nutrisense acquisition contributed incremental non-CGM revenue that management sized at “the millions, you can count on one hand,” and most of what flowed through Nutrisense was already Dexcom Stelo revenue. Organic growth of 12% versus 13.1% reported is the FX and small-acquisition wedge, not an acquisition engine.

Where the growth comes from now. Four distinct engines, in descending order of current contribution:

1. International access expansion (16% organic). Dexcom is still winning first-time Type 1 coverage in several of its top-ten ex-US markets, before basal and non-insulin are even addressed. Management sizes the core international opportunity at “well north of 60 million potential lives.” The mechanism is tenders: win reimbursement, win the tender, take share. Flex and ONE+ exist specifically to compete in tenders Dexcom previously could not enter on price. The near-term caveat management flagged is a tougher Q3/Q4 comparison and an FX drag of roughly $15M in the second half.
2. US under-penetration of already-covered lives. ~9 million Americans have coverage and no CGM. This is the cheapest growth available — no payer negotiation, no new clearance, just detailing and prior-authorisation removal.
3. The 15-day conversion. Roughly 50% of the US base by year-end 2026, from a full-year average of 20–25%. Because it starts 2027 at ~50% and rises, the larger revenue-per-patient and margin benefit lands in 2027, not 2026 — management was explicit that “it really starts to contribute next year.” Internationally the same conversion has barely begun; Canada is the first clearance.
4. CMS non-insulin Type 2. The step-function. A decision is expected before year-end 2026 with coverage modelled from mid-2027 across ~25 million Americans.

Beyond those, the longer-dated optionality is G8 (2027 submission, end-2027/early-2028 launch, half the size, multi-analyte with ketones and potassium), Smart Basal (basal-insulin titration, addressing the >70% of basal patients who fail to reach target A1C within a year), Stelo/prediabetes (115 million Americans, most undiagnosed, now with a rebuilt app and pediatric clearance and a TEMPO pathway for screening claims), and the Nutrisense-derived nutrition-coaching layer that management intends to push across the entire portfolio.

Growth quality — the honest assessment. The quality is improving but is not uniformly high. High-quality attributes: it is organic, it is recurring, it comes with expanding rather than contracting margins, and the incremental operating margin of 67.6% shows it is genuinely leverageable. Lower-quality attributes: the fastest-growing pools (international tender business, non-insulin Type 2, OTC) carry lower ASP and weaker switching costs than the shrinking-share-of-mix intensive-insulin core, and a material portion of the 2027 growth case depends on a regulatory decision Dexcom does not control. A growth algorithm with a binary government catalyst embedded in it is lower-quality than one built purely on share and price, however attractive the expected value.

Verdict (Growth): Good-quality growth, decelerating in rate and improving in economics. Low-teens organic growth is durable and well-diversified across four engines, and it is now converting to earnings at roughly 3.5x the top-line rate. The rate is materially below Dexcom’s history and the 10%+ LRP floor is a realistic rather than heroic commitment. The reservation is dependency: the difference between a 10% and a 13% grower in 2027–28 is substantially a CMS decision, and underwriting the upper half of that range is underwriting a regulator.

6. Financial Quality

This is where the update is most emphatic. On the metrics that separate a good business from a story, DexCom’s trajectory over five quarters is one of the cleaner recoveries in large-cap medtech.

The margin and return series. Trailing-twelve-month, period-end:

ROIC has doubled in five quarters, from 9.5% to 19.2%. That is the single most important number in this report. A 19.2% return on invested capital is comfortably above any plausible cost of capital for a net-cash, 0.85-beta business, and it means growth is now unambiguously value-creating — which was genuinely arguable at 9.5%. ROE of 38.5% is flattered by buyback-shrunk equity ($2,621M against $6,455M of assets), so ROIC and ROA are the honest measures; both are strong and both are rising steeply.

Is the margin recovery real or borrowed? I pressure-tested three ways it could be artificial. (a) Price. No — management attributes the gain to scrap, throughput and quality management, and Dexcom is simultaneously pushing into lower-ASP tiers; if this were price, mix would be fighting it, not amplifying it. (b) Under-investment. Partially worth watching — capex fell to $161.3M in H1 2026 from $181.1M, and R&D was $153.0M in Q2 versus $148.2M, up only 3.2% against 13.1% revenue growth. R&D at 11.7% of revenue is still healthy for the category, and the G8/Smart Basal/app pipeline is visibly funded, so I do not read this as starvation — but a second year of sub-inflation R&D growth would change that. © Deferred cost. The Ireland factory start-up is a real cost about to arrive: management stated explicitly that gross margin peaks in Q3 and steps down as Ireland turns on, because a newly started factory carries very high per-unit cost before it leverages. The implied second-half gross margin is roughly in line with the first half despite the growing 15-day tailwind — that is the Ireland offset, disclosed in advance. I regard this as good disclosure rather than a negative surprise, but it does mean the reported margin trajectory flattens from here through 2027 before resuming.

Cash generation. H1 2026 operating cash flow of $794.8M versus $486.8M, with capex of $161.3M versus $181.1M, produces free cash flow of $633.5M versus $305.7M — a 107% increase on 12% revenue growth. On a trailing basis, ROIC.ai computes TTM free cash flow to the firm of ~$1.42B. FCF conversion of GAAP net income is running comfortably above 100%, which is what you expect from a business whose D&A ($68.2M in Q2) now exceeds its quarterly capex run-rate. That last observation is also the caution: FCF is currently being flattered by capex running below depreciation, and that cannot persist indefinitely in a company building factories in Ireland and Malaysia.

Balance sheet. Cash and equivalents $1,105.0M plus short-term marketable securities $842.0M equals $1,947.0M of liquidity, against total debt of $1,398.4M — principally the $1.25B 2028 convertible notes carried at $1,242.8M. Net cash of approximately $549M. The company is in compliance with the leverage and fixed-charge covenants on its amended credit agreement. The convertible notes remain a dilution tail: they can convert into cash and/or shares before the May 2028 maturity if conditions are met, and the diluted share count (390.1M) exceeds basic (382.2M) by roughly 2.1% largely on that account, partially offset by capped calls.

Working capital — a modest watch item. Inventory rose from $629.1M at year-end to $726.4M, up 15.5% against 12% first-half revenue growth. Receivables rose only 3.4% to $1,257.7M, well below revenue growth, which is the reassuring half of the read — Dexcom is not stretching terms to make the quarter. The inventory build is consistent with the 15-day ramp and the Ireland start-up (you build stock before you switch lines), and management’s scrap commentary argues the inventory is good rather than obsolete. I flag it as a quality-of-earnings item to re-check at Q3, not as a red flag.

Dilution and share-based compensation. SBC was $39.0M in Q2 2026 versus $45.3M a year earlier — falling in absolute dollars while revenue grew 13% — and $82.3M for the half. At roughly 3.2% of revenue and declining, SBC is moderate for the sector and is more than fully offset by buybacks: the $600M repurchase in Q2 alone is more than seven times the half-year SBC charge. Unrecognised RSU/PSU cost is $263.6M over 2.2 years, so the run-rate is stable. Share count is falling, not rising.

Tax. The estimated FY2026 effective rate is 21.8%; the H1 actual was 23.6%, elevated by share-based-compensation shortfalls and disallowed executive compensation. OECD Pillar Two is not currently material. The California Franchise Tax Board opened an audit of the 2022 and 2023 California returns in June 2026; management does not expect significant adjustments. None of this is thesis-relevant, but the 180bp gap between guided and actual rate is a modest drag on reported EPS in the near term.

One-time items and run-rate distortion. I checked the multi-year set for items flattering the run-rate. The one that stands out is in the comparison base, not the current period: Q3 2025 GAAP net income of $283.8M exceeded operating income of $242.5M because of an $81.8M non-operating gain. Any analysis anchoring on trailing GAAP net income therefore embeds a non-recurring benefit; this is why I lean on operating income, EBITDA and ROIC rather than GAAP net income for the valuation work below. There is no comparable distortion in Q1 or Q2 2026.

Verdict (Financial Quality): Economics improve materially with scale, and the proof is now in the numbers rather than the guidance. Six consecutive quarters of trailing margin expansion, a doubling of ROIC to 19.2%, a 67.6% incremental operating margin, free cash flow more than doubling on 12% revenue growth, net cash, falling SBC and a shrinking share count — this is a high-quality financial profile and a genuine inflection rather than a rebound. Two honest qualifications: the near-term margin path flattens as the Ireland factory turns on, and current free cash flow is flattered by capex running below depreciation. Neither undermines the verdict; both argue against extrapolating the rate of improvement.

7. Capital Allocation

My earlier report graded capital allocation B+, praising the organic discipline and net-cash balance sheet while criticising pro-cyclical buyback timing, an in-the-money convertible, a growth-weighted incentive plan and sub-1% insider ownership. Three of those five judgements need revision.

Buybacks — the criticism is now falsified by the record. In May 2026 the board terminated the 2025 $750M program with $250M unused and authorised a $1.00 billion 2026 program running to June 30, 2027. In Q2 alone Dexcom repurchased 8.6 million shares for $600.0 million, an average of approximately $69.77 per share — the largest quarterly repurchase in the company’s history, executed into the trough of the year and roughly 21% below the current $84.75 price. Set against the earlier report’s complaint that Dexcom “bought more in 2024 at higher prices and less in 2025 as the stock fell,” this is the opposite behaviour: maximum deployment at minimum price. Cumulative repurchases now run to roughly $3.1B since 2022 ($557.7M, $688.7M, $750M, $500M, and $600M year-to-date 2026).

A capital-return policy now exists, for the first time. The 10-Q states Dexcom expects to allocate “at least 50% of cash generated from operations, net of planned capital expenditures, to share repurchases over the course of our long-range plan.” Against H1 free cash flow of $633.5M, a ≥50% commitment implies a baseline of roughly $650–750M per year at current run-rates, scaling with FCF toward the 2030 margin targets. This converts buybacks from opportunistic to structural and is the most shareholder-friendly disclosure Dexcom has ever made. There is still no dividend, which I regard as correct for a business reinvesting at a 19.2% ROIC.

M&A — the organic record ends, gently. Dexcom acquired Nutrisense in June 2026 for $41.2M net of cash acquired — 0.13% of market capitalisation. Goodwill was essentially unchanged at $24.1M while intangibles rose from $70.8M to $101.6M, meaning the consideration was allocated almost entirely to identifiable technology rather than goodwill. Management’s rationale is explicitly technological — a CGM-guided nutrition-insight and coaching platform, from a long-standing partner, whose engagement scores management cites as unusually high — and they were candid that incremental non-CGM revenue is trivially small because most Nutrisense revenue was already Dexcom Stelo pass-through. (FACT.) Interpretation: this is a well-priced, well-understood tuck-in of a known partner, and it is the right kind of deal. But the Investor Day also established a stated appetite for “transactions that have the potential to accelerate our technology pipeline,” which is a new risk surface. The earlier report’s clean “no acquisition history, no integration risk” characterisation no longer holds in full, and future deals deserve scrutiny that this one does not require.

Capital expenditure. Capex of $161.3M in H1 2026 (down from $181.1M) funds Arizona, Malaysia and the Ireland facility that begins commercial production later in 2026. At roughly 6.5% of revenue annualised — down from ~8% — this is a moderately capital-intensive business whose intensity is currently falling. Management is investing ahead of demand in Ireland (hiring and training a workforce a quarter before lines start, with the associated depreciation), which is the correct sequence and is being disclosed transparently as a near-term margin cost.

Incentives and alignment — largely unchanged, and still the weakest link. Say-on-pay passed at the May 27, 2026 annual meeting with 277,111,475 for against 36,664,959 opposed — 88.3% support of votes cast. That is passing but unremarkable; well-run large caps with uncontroversial pay routinely clear 95%. Director Nicholas Augustinos drew 47,957,634 against votes (84.7% support), the highest opposition on the board. Insider ownership remains well under 1% of the company. The board was fully re-elected including all twelve nominees.

Insider behaviour — uniformly one-directional, and worth stating plainly. Every Form 4 filed between June 1 and August 4, 2026 was a sale. Former CEO and current director Kevin Sayer sold 26,756 shares on July 6 at $72.00 and again on July 20 at $76.32, leaving 328,970 held directly. Director Mark Foletta sold 2,000 on July 31 at $80.00 and 2,000 on August 3 at $85.32 through a trust. Officers Michael Brown and Jon Coleman made comparable small, regular sales. All carry the footnote denoting pre-arranged Rule 10b5-1 plans, so none is a discretionary conviction signal. (FACT.) Interpretation: 10b5-1 selling is diversification and carries little information. What carries some information is the asymmetry with the prior period: the earlier report highlighted the incoming CEO buying roughly $1M of stock in the open market at the $55 lows. There is no equivalent buying at $85. That is entirely rational behaviour and is not a bearish signal — but it does mean the “insiders are buying” leg of the June thesis has expired.

Verdict (Capital Allocation): Upgrade to A−. The single largest criticism in the earlier report — pro-cyclical repurchase timing — has been decisively rebutted by a record $600M deployed at ~$69.77 into the year’s trough. A formal ≥50%-of-FCF return policy now exists. M&A opened, but at 0.13% of market cap with no goodwill and a clear technology rationale. Capex is disciplined and disclosed. What holds the grade below an A is unchanged: sub-1% insider ownership, an 88% say-on-pay vote that signals residual shareholder scepticism about incentive design, and a convertible dilution tail into May 2028.

8. Changes and Headwinds — Last Two Years

Timeline of material events.

The quality-system overhang — status changed, but not resolved. The earlier report identified the March-2025 FDA warning letter as the most underappreciated risk, escalated specifically because the FDA judged Dexcom’s responses to earlier Form 483 observations inadequate, and layered with a Class I receiver recall and May-2026 Class II app recalls into what looked like a systemic pattern. The Q2 2026 10-Q does not mention the warning letter, Form 483 observations, or a consent decree anywhere in its text. (FACT.) Interpretation — and I want to be precise here, because it cuts both ways. Silence is not closure: a closed warning letter is normally announced by the company, and no re-inspection or close-out letter is on the public record. But it is also not nothing — a company that believed a consent decree was a live possibility would ordinarily be discussing it in a risk narrative, and management on the Q2 call spent time on the operational side of the same story, reporting that complaint levels have come down, out-of-box failure rates have been addressed, NPS has risen three consecutive quarters, and scrap reduction is a principal driver of the gross-margin recovery. That is exactly the operational evidence you would expect if the underlying quality-system deficiencies were being remediated successfully. I downgrade the probability of escalation but keep the item open, and it remains my second-listed bear trigger.

Litigation — real but unquantified. The 10-Q discloses a securities class action, derivative actions, and G6/G7 product class actions, and states plainly that Dexcom is “unable to reasonably estimate the ultimate outcome of any of the litigation matters at this time.” No loss accrual is disclosed. The securities complaint covers purchasers between January 8, 2024 and September 17, 2025 — a period bracketing the July 2024 guidance cut and the 2025 margin trough — so the alleged wrongdoing concerns disclosure during the very events this report treats as historical. Two law firms publicised investigations in July 2026. (FACT.) Interpretation: securities class actions following a −40% single-day guidance cut are close to automatic and usually settle within the D&O tower; the tail risk is modest but the absence of any accrual means the eventual charge, whenever it comes, will be a surprise to the P&L rather than a release from a reserve.

A minor but telling operational event. In June 2026 Reuters reported that Dexcom sensors slated for destruction had been stolen during disposal and resold into the market through a licensed wholesaler chain. Financially immaterial; operationally it is a reverse-logistics control failure, and it fits the same family as the quality-system observations. Small, but not nothing, for a company whose central credibility question is process discipline.

Verdict (Changes & Headwinds): The balance has tipped clearly positive for the first time since mid-2024. Over the trailing two years the ledger was dominated by self-inflicted damage. Over the trailing three months it is dominated by execution: a published long-range plan, a record buyback at the low, Level-A clinical evidence in the largest untapped segment, the first ex-US 15-day clearance, an FDA pilot selection, and a beat-and-raise. The negatives that remain — an unresolved-on-the-record warning letter, unaccrued litigation, a 2028 competitive-bidding price-down — are real and are why a discount to a best-in-class compounder multiple is still warranted. But the direction of travel reversed, and the market has correctly noticed.

9. Risk Analysis

The three that actually matter. First, valuation — this is now the largest risk in the name, which was not true in June. Second, the CMS decision, which is binary, dated within five months, and worth several points of 2027–28 growth in either direction. Third, whether the margin recovery is a level shift or a trend; the Ireland step-down means the next two or three quarters will not show the clean expansion the last six did, and the market has not obviously priced a pause.

What has moved down the list. The FDA warning letter, which the earlier report ranked as the most underappreciated overhang, is now my sixth-listed risk: the operational evidence (falling complaint rates, falling scrap, rising NPS) is consistent with successful remediation, and the item’s absence from the Q2 10-Q — while not proof of closure — is not the behaviour of a company bracing for a consent decree. The GLP-1 demand risk, prominent in the 2024–25 bear case, is close to falsified by CONNECT and no longer merits a line in the matrix.

10. Valuation Discussion — Embedded Expectations

Where the price is. At the August 7, 2026 close of $84.75, with approximately 377.3 million shares outstanding (385,872,977 at the April 1 record date less 8.6 million repurchased), market capitalisation is roughly $32.0B, or $33.1B on the 390.1 million diluted share count. Net cash of approximately $549M — $1,947.0M of cash and short-term securities less $1,398.4M of total debt — gives an enterprise value of roughly $31.4B.

Multiples, trailing and forward.

Own-history percentile — the central valuation change. Against its own approximately ten-year range, DXCM’s composite valuation percentile is 9.8 — P/B at the 9.6th percentile and P/S at the 11.6th. (FACT — AZI valuation index, pulled 2026-08-07.) Data caveat, stated plainly: the published P/E percentile of 8.3 rests on a trailing EPS of $2.314, which is the window ending one quarter early (Q2’25 $0.4405 + Q3’25 $0.6966 + Q4’25 $0.6701 + Q1’26 $0.5069 = $2.3141 exactly). The correct trailing four quarters sum to $2.5122, so the true trailing P/E is 33.7x rather than the published 36.6x and the true percentile is somewhat below 8.3. The P/B and P/S legs are unaffected and are the reliable reads. The direction is unambiguous either way.

The comparison that matters is with ourselves two months ago. The earlier report found DXCM in the bottom ~3rd percentile of its own ten-year range. It now sits in roughly the bottom 10th. Roughly a third of the valuation gap has closed in two months. The stock is still historically cheap against its own history — but “cheapest in a decade” and “cheaper than 90% of the last decade” are materially different propositions, particularly when the last decade included a period at 30x sales that no rational analyst would use as an anchor.

Embedded expectations — what the price requires. This is the heart of the update. An enterprise value of $31.4B against trailing EBIT of $1,139M is 27.6x. On a 10% discount rate and 3% terminal growth, a 27.6x trailing-EBIT multiple requires operating income to roughly double over five to seven years to deliver an adequate return. Test that against management’s own long-range plan:

The LRP: organic revenue growth above 10% in every year through 2030, a 2030 non-GAAP operating margin of 29–30%, gross margin in the high-60s, and adjusted EBITDA margin of 36–37% — with competitive bidding assumed from 2028.

Compounding FY2026 guided revenue of roughly $5.2B at 10–11% for four years gives 2030 revenue of approximately $8.3–8.5B. At a 29.5% operating margin that is $2.45–2.50B of operating income, against $1,139M trailing — a 2.15x increase. The plan, delivered in full, produces almost exactly the earnings growth the current multiple requires.

That is the finding. At $57.82 on Investor Day, and to a lesser degree at $76.62 in June, the long-range plan was free optionality layered on top of a cheap stock. At $84.75 it is the base case. The stock has migrated from “cheap if management does nothing special” to “fairly valued if management executes an ambitious plan through a scheduled 2028 government price-down.” That is a fundamentally different risk profile at the same business quality.

Scenario analysis. Assumptions are stated; none is a forecast.

*Per-share figures assume the ≥50%-of-FCF buyback policy modestly reduces share count over the period and add back net cash; they are illustrative arithmetic, not targets, and appear here solely to size the scenario spread.

Read the distribution honestly. The spread is still positively skewed — roughly flat in the bear case against +50% in the base and +95% in the bull. But note what changed: the bear case is now approximately today’s price. In June, with the stock at $76.62, the same bear arithmetic implied roughly +11%; at $57.82 in May it implied +47%. Downside protection has been consumed by the rally. An investor buying today is underwriting the plan, not buying an option on it.

Cross-check: what would justify a higher multiple? Three things, and they are worth naming because two are plausible. (1) A durably higher terminal margin. If the 2030 gross-margin target of high-60s proves conservative because G8 is cheaper to build than modelled, the operating-margin path could exceed 30%. (2) A CMS decision that expands the 2030 revenue base above $9B, which the bull case captures. (3) Multiple re-rating toward the medtech-compounder cohort — Dexcom at 22.4x trailing EBITDA against a high-quality medtech cohort that has historically cleared 25–30x. That third argument is real, but it is precisely the argument that has already been half-made by the market since May, and re-rating arguments are the weakest kind of valuation support.

Sum-of-the-parts. Not warranted. Dexcom is a single-segment business; Stelo and Flex are product tiers within one manufacturing and commercial platform, not separable assets. Nutrisense at $41.2M does not move a $32B capitalisation.

Verdict (Valuation): Fair, not cheap — and the change in that assessment is the reason this report exists. The business is demonstrably higher-quality than it was in June, and its multiple remains low against its own history. But the arithmetic at 27.6x trailing EBIT now requires the full delivery of an ambitious five-year plan through a scheduled regulatory price-down, with a binary CMS catalyst pending inside five months. The margin of safety that defined the June setup has been substantially consumed. No price target or recommendation appears in this section; the single opinion in this memo is in Claude’s Take.

11. Variant Perception

What consensus now believes. The consensus has flipped in three months from sceptical to constructive. Sell-side coverage following the Q2 print reads as “beat, raise, margins inflecting, buy the recovery”; FY2026 EPS estimates sit at $2.57 and FY2027 at $3.05, implying 23% and 19% growth. The stock made a 52-week high four days ago and has outperformed every large medtech peer year-to-date — Dexcom +27.7% against Abbott −12.4%, Intuitive −33.1% and Insulet −50.3%. Consensus is now positioned for the margin recovery to continue and for CMS to eventually expand non-insulin coverage.

The strongest bull case. Dexcom is an early-innings compounder in a category with a 25-million-patient US TAM about to be unlocked by a regulator, run by a management team that has just published a credible plan and backed it with the largest buyback in company history at the year’s low. Five supports: (1) six quarters of margin expansion driven by process, not price, with a 67.6% incremental operating margin; (2) ROIC doubling to 19.2%, which makes growth unambiguously value-creating; (3) the 15-day conversion is only ~20–25% averaged across 2026 and starts 2027 near 50% and rising internationally from a base of essentially zero — the largest margin and revenue-per-patient benefit is still ahead; (4) CONNECT gives Level-A randomised evidence in the largest untapped segment, and the GLP-1 subgroup result converts the principal bear narrative into a tailwind; (5) a ≥50%-of-FCF buyback policy compounding per-share value on a shrinking count.

The strongest bear case. The bear case is no longer about demand; it is about price and duration. (1) At 27.6x trailing EBIT the market has already capitalised the 2030 plan, so the reward for success is ordinary and the penalty for shortfall is large. (2) The margin curve flattens near-term by management’s own guidance as Ireland turns on — a market that has re-rated on margin momentum will not enjoy two or three quarters of no margin momentum. (3) Growth is low-teens and decelerating, with tougher second-half international comparisons and a disclosed FX drag; the LRP floor is “above 10%,” not low-teens, which is management telling you the honest number. (4) Competitive bidding arrives in 2028 — a government-administered price-down in the largest reimbursed channel is not a risk the company can execute its way around, only absorb. (5) Free cash flow is currently flattered by capex running below depreciation, so the 4.0% FCF yield overstates the sustainable figure. (6) The unresolved warning letter and unaccrued litigation are both unquantified negative options.

The three to five assumptions that actually decide it.

1. Does gross margin resume expanding in 2027 after the Ireland pause? The entire multiple rests on the 2030 high-60s gross-margin target. The 15-day conversion mathematics (US base at ~50% entering 2027 and rising, international barely started, G8 designed cheaper) says yes. Ireland says not for two or three quarters. This is the highest-frequency, most observable variable.
2. Does CMS extend non-insulin Type 2 coverage, and when? Decision expected before year-end 2026, coverage modelled mid-2027. Binary, dated, and worth several points of 2027–28 growth.
3. How much does competitive bidding take in 2028? Management says the 10%+ growth floor holds through it. Nobody outside CMS knows the magnitude.
4. Is the 2024 execution failure genuinely behind them? Rising NPS three quarters running, falling complaint rates, falling scrap and a record buyback at the low all say yes. A 27.6x multiple requires it to stay yes.
5. Does Abbott choose to compete on price in non-insulin Type 2? Dexcom expanded gross margin 400bp while pushing into that segment, which says not yet.

What the tape and the factor model say about positioning. DexCom’s factor profile argues this is a stock-specific outcome, not a factor trade. Market beta is 0.79 and industry (Medical Devices) beta 1.40 in the Base+Sector+Industry model, but model R² is only 0.264 and idiosyncratic volatility is 37.6% annualised — roughly three-quarters of return variance is company-specific. The empirical proof is in this year’s tape: Dexcom is +27.7% year-to-date while Abbott, Intuitive and Insulet are all sharply negative. This is not a medtech rotation; the sector is out of favour and Dexcom is being rewarded for its own execution.

The risk-adjusted track record locates the stock precisely. Three-month annualised return is +282% with a Sharpe of 5.63 — a figure that must be read correctly: every FactorsToday horizon is annualised, so this is one exceptional quarter (+39.8% raw, reconciling exactly to the price series) extrapolated, not a durable Sharpe. Six-month annualised is +46.7% at a Sharpe of 1.03. But three-year annualised is −11.5% and five-year is −7.9%, with a three-year maximum drawdown of −60.9%. The long-horizon record is still that of a value name that broke, and relative strength versus its own peak remains −48.0%.

My variant perception, stated plainly. In June the variant view was that the market was wrong to treat a transient margin trough as structural. That variant is now consensus and has been paid. My current variant is narrower and less comfortable: the market is now extrapolating the rate of margin improvement into a period management has explicitly guided will pause. Six consecutive quarters of trailing margin expansion have trained the market to expect a seventh, and the Q3 print will likely deliver the peak while Q4 and 1H27 absorb the Ireland start-up. If the multiple was awarded for momentum in margin, a flat margin quarter is a re-rating risk even with the long-term plan fully intact. The offsetting variant, in the other direction, is that the market may be under-crediting the 2027 step-up in 15-day economics, which management describes as where the transition “really starts to contribute” and which is largely invisible in FY2026 numbers. Those two variants roughly offset — which is precisely why my position on the name is neutral rather than directional.

Where consensus might be offsides. Consensus appears to be treating the CMS non-insulin decision as a probable positive already embedded in estimates. It is not embedded in the LRP’s 10% floor — management built the plan without it and models it separately from mid-2027. If that is right, a favourable CMS decision is genuine upside to the plan rather than a component of it, and the bull scenario is more accessible than the base case implies. Conversely, if consensus estimates have quietly capitalised CMS, an adverse or delayed decision would be a larger negative surprise than the price suggests. This asymmetry is the single most useful thing to resolve before adding to a position.

12. Fact vs. Interpretation

13. Open Questions

1. Is the March-2025 FDA warning letter closed? The Q2 2026 10-Q does not mention it, and there is no re-inspection or close-out letter on the public record. Silence is ambiguous. This is the cleanest question to put to investor relations, and the answer materially affects the tail risk.
2. Will CMS extend national coverage to non-insulin Type 2, and on what terms? Management expects a decision before year-end 2026 and has modelled mid-2027 effectiveness. Terms matter as much as the yes/no — an A1C-threshold restriction, which management argues would be inconsistent with both CMS precedent and existing commercial coverage, would materially shrink the unlock.
3. How much does competitive bidding take from ASP starting 2028? Management says the 10%+ organic growth floor survives it. The magnitude is unknown and is the largest unquantified structural headwind in the model.
4. Is the H1 capex decline to $161.3M a timing artefact or a durable step-down? Capex is running below depreciation. FY2027 capital guidance will answer it, and it matters because roughly half of the free-cash-flow improvement is coming from the investment line rather than the margin line.
5. What is the actual 2027 gross-margin bridge? Management has guided a Q3 peak and an Ireland step-down but has not quantified the 2027 path between ~64% and the 2030 high-60s target. Without that bridge, the terminal margin assumption is management’s word.
6. What is the Stelo installed base and its retention curve? Dexcom still does not disclose Stelo units, revenue or repeat-purchase rates. Without them the OTC/prediabetes leg of the story is unfalsifiable.
7. What is Dexcom’s tuck-in M&A appetite in dollar terms? Nutrisense at $41.2M is trivial, but the Investor Day framework explicitly opened the door to technology-pipeline acquisitions. A capital-allocation framework that is 50% buyback leaves the other 50% undefined.
8. What is the expected loss range on the securities class action? No accrual is disclosed and the company states it cannot reasonably estimate the outcome. The eventual charge will therefore be a surprise rather than a reserve release.
9. Do the AID partners have any exclusivity horizon at all? Every partner supports Abbott. If there is no path back to preferential integration, the captivity leg of the moat erodes structurally with each new Abbott integration.

14. What Must Be True

For the bull case

1. Gross margin resumes expanding in 2027 after the Ireland pause, tracking toward the high-60s 2030 target on 15-day conversion (US base near 50% entering 2027 and rising; international barely begun) and G8’s lower manufacturing cost.

   Falsification test: FY2027 guidance issued with the Q4 2026 print implies a non-GAAP gross margin below ~65%, or Q1/Q2 2027 non-GAAP gross margin prints below the ~64% FY2026 exit rate. Either would break the terminal-margin assumption that carries the multiple.

2. CMS extends non-insulin Type 2 coverage on broad terms, adding a materially larger reimbursed population from mid-2027.

   Falsification test: CMS declines, defers beyond 2027, or imposes an A1C threshold or step-therapy restriction that limits eligibility to a minority of the ~25 million population.

3. Organic growth holds at or above 10% through the 2028 competitive-bidding transition, as the LRP asserts.

   Falsification test: any quarter of organic growth below 8%, or FY2028 revenue guidance implying single-digit growth once bidding rates are published.

4. The execution repair is durable — NPS, complaint rates and scrap continue improving, and the sales force converts the ~9 million covered-but-unused population.

   Falsification test: US revenue growth decelerating below high-single digits without a coverage change, or a sequential decline in new-patient starts for two consecutive quarters.

5. Capital returns continue at the stated policy, with at least 50% of free cash flow to buybacks and no large, poorly-priced acquisition.

   Falsification test: an acquisition above roughly $1B, or repurchases falling below the stated 50% threshold in any full year of the plan.

For the bear case

1. The margin recovery was a level shift, not a trend — the easy scrap, freight and quality-remediation gains are banked, and the next 500bp to the 2030 target proves far harder than the last 400bp.

   Falsification test: two consecutive quarters of non-GAAP gross margin above 65% after the Ireland start-up is absorbed (i.e. in 2027), which would demonstrate the curve is still bending.

2. Competitive bidding from 2028 compresses ASP more than the LRP budgets, converting a 10%+ grower into a high-single-digit grower at exactly the moment the multiple demands the opposite.

   Falsification test: published CMS bidding rates that imply a mid-single-digit or smaller ASP reduction, with management reaffirming the 10%+ floor alongside them.

3. Abbott escalates on price in Type 2 non-insulin and OTC, where switching costs are weakest and where both firms are heading.

   Falsification test: Dexcom holding or expanding gross margin over four consecutive quarters while non-insulin mix rises — which is, notably, exactly what the last four quarters showed.

4. The valuation is the risk — at 27.6x trailing EBIT the plan is priced, so a plan delivered on time earns an ordinary return while any shortfall re-rates the stock sharply.

   Falsification test: a CMS approval or a G8 development that credibly lifts 2030 revenue above ~$9B, which would make today’s multiple cheap against a larger base rather than fair against the planned one.

5. The quality-system pattern reasserts itself — the warning letter escalates, or a further Class I recall lands on a device patients dose insulin from.

   Falsification test: a published FDA close-out letter for the March-2025 warning letter, or a clean re-inspection with no new Form 483 observations.

The single most informative near-term datum is the CMS non-insulin decision, expected before year-end 2026. The single most informative recurring datum is the FY2027 gross-margin guide issued with the Q4 2026 print — it is the first point at which management must commit, in numbers, to the glidepath between today’s ~64% and the 2030 high-60s target that the current multiple already assumes.

15. Source Appendix

See Appendix B below for the full list of primary filings, transcripts, data feeds and secondary sources relied upon, with URLs and access dates.

The analysis above contains no investment recommendation and no price target; the only position taken in this article is in the clearly-labelled Claude's Take block at the top, which is the author’s own subjective view. This article is general information and is not investment advice. The author has no position in DexCom, Inc. and no business relationship with any company mentioned.

APPENDIX A — Standard Diligence Questionnaire

DexCom, Inc. (NASDAQ: DXCM) · August 8, 2026

A supplemental due-diligence questionnaire covering the standard checklist. Answers are labelled FACT / INTERPRETATION / ASSUMPTION where the distinction matters.

General

What thoughtful questions have other investors asked about this company?

The sell-side Q&A on the Q2 2026 call clustered around six themes, and they are a fair map of what serious investors currently care about. (1) Is the US CGM market’s ~10% growth rate sustainable, and what happens to it when coverage expands? — management’s answer was the ~9 million covered-but-unused Americans, i.e. growth before any coverage event. (2) What is the realistic CMS timeline for non-insulin Type 2, and could it be sooner than mid-2027? — no change to assumptions; decision expected before year-end 2026. (3) Where does the G7 15-Day conversion sit, and when is the peak margin benefit? — roughly 20–25% averaged across 2026, approaching 50% at year-end, with the material contribution landing in 2027 and continuing as international converts. (4) Is the operating leverage sustainable or a timing artefact? — management flagged Ireland hiring and depreciation as a near-term offset while asserting the underlying levers persist. (5) Is the gross-margin raise underlying performance or a softer oil/resin assumption? — the CFO attributed the raise to line throughput and efficiency, with the resin/freight headwind tracking as previously guided. (6) Can you quantify product-performance improvement? — NPS up three consecutive quarters, complaint levels down, out-of-box failure rates addressed, and scrap reduction visible directly in the gross margin.

The question that was conspicuously not asked on the call: the status of the March-2025 FDA warning letter. That silence is itself informative about where the market’s attention has moved. (INTERPRETATION.)

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? Neither in the classic sense — Dexcom is a secular-growth medical-device business, not a cyclical. But earnings are at a margin-cycle inflection, coming off a trough rather than sitting at a peak. Trailing operating margin bottomed at 15.25% in the twelve months to March 2025 and is now 22.92%, against a 2030 target of 29–30%. On the company’s own framework, earnings are roughly two-thirds of the way from trough to plan. (FACT on the series; INTERPRETATION on the positioning.)

Driven by the external environment or internal actions? Overwhelmingly internal. The 2024–25 collapse was self-inflicted (a sales-force reorganisation colliding with the DME-to-pharmacy channel shift) and the recovery is likewise self-generated: scrap reduction, manufacturing throughput, quality management, the 15-day mix, and operating-expense discipline. The external environment — reimbursement expansion, GLP-1 adoption, category growth — was broadly favourable throughout both the fall and the recovery. That is a strong point in management’s favour on the way up and an indictment on the way down. (INTERPRETATION.)

How stable are revenues? Very. The model is recurring sensor reorders from an installed patient base, with revenue recognised on shipment to distributors. There has been no year of revenue decline in the company’s modern history, and the 2024 crisis was a growth-rate disappointment, not a revenue decline. Quarter-to-quarter variability comes from channel inventory timing, international tender timing and FX, not from demand volatility.

Outlook for products/services? The product ladder is the strongest it has ever been: G7 15 Day at the premium end (MARD 8.0%, 15.5-day wear), ONE+ and the new Flex for international value tiers and Type 2 basal/non-insulin, Stelo over the counter including pediatric, Smart Basal as a dosing-software layer, and G8 in 2027–28 at half the size with multi-analyte capability. The strategic shift from a premium-only position to a full price ladder is the right structural answer to a volume competitor. (FACT on the portfolio; INTERPRETATION on the strategy.)

How big will this market be — growing, shrinking, domestic or international? Growing, and international is the faster half. The global CGM market is roughly $11–13B growing ~15% annually toward the early 2030s; sensors are ~84% of value. Dexcom’s international revenue grew 19% reported / 16% organic in Q2 versus 11% in the US and is now 28.7% of the total. Management sizes the core international opportunity at “well north of 60 million potential lives” across its top-ten ex-US markets, with Type 1 coverage still being won in several of them before basal and non-insulin are addressed. Domestically the unlock is the ~25 million non-insulin Type 2 population pending CMS and the ~115 million prediabetes population addressed by Stelo.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? Roughly stable at a high level, with the intensity shifting downmarket. It remains a Dexcom/Abbott duopoly. The competitive front has moved from the intensive-insulin tier (where Dexcom’s AID integration and accuracy defend a premium) to Type 2 non-insulin and OTC (where price matters more and switching costs are weakest). The observable evidence that a price war is not currently underway is that Dexcom expanded gross margin 400bp year-over-year while simultaneously pushing into the lower-ASP tiers. The scheduled 2028 arrival of CMS competitive bidding will raise intensity structurally regardless of what the two firms choose. (FACT/INTERPRETATION.)

How profitable is the business (ROIC, ROE)? Highly, and improving fast. Trailing twelve months to June 30, 2026: ROIC 19.21%, ROE 38.49%, ROA 14.51%, return on capital 22.05%, gross margin 62.47%, operating margin 22.92%, net margin 20.12%. ROIC has doubled from 9.48% in the twelve months to March 2025. ROE is flattered by buyback-shrunk equity ($2,621M against $6,455M of assets), so ROIC and ROA are the honest measures. Both comfortably exceed any plausible cost of capital for a net-cash, 0.85-beta business. (FACT — ROIC.ai, reconciled to filings.)

How profitable is the industry — how many competitors, what barriers to entry? Two serious competitors plus three minor ones (Medtronic, Senseonics/Ascensia, Roche). Barriers are genuine: FDA/CE clearance for a device patients dose insulin from; accuracy validated in registrational trials; automated high-volume manufacturing with scrap rates that only cumulative volume teaches; payer contracting and formulary placement; and AID ecosystem integration. Medtronic’s decision to source an Abbott-built sensor for the 780G is the strongest available evidence of how hard these barriers are — a $109B medical-device company chose to buy from a competitor rather than build. (FACT/INTERPRETATION.)

Can the business be easily understood? Yes. It sells a disposable sensor that people with diabetes replace every 10 to 15 days. Revenue is units times price; margin is price minus a manufacturing cost that falls with volume and with wear duration. There is no financial engineering, one segment, and a clean balance sheet.

Can it be undermined by foreign low-cost labour? Not meaningfully. The cost structure is automated capital equipment, sensor chemistry and yield, not labour. Dexcom manufactures in the US (San Diego, Mesa/Arizona), Malaysia and now Ireland; the offshore facilities exist for capacity, tariff and proximity-to-market reasons rather than wage arbitrage. Regulatory clearance and clinical validation, not labour cost, are what a low-cost entrant cannot buy. (INTERPRETATION.)

Do brands matter? Modestly, and more than in most medtech. “Dexcom” carries genuine recognition among endocrinologists and in patient communities, and NPS is a metric management now reports movement on. But brand is not the moat — an equally accurate, better-reimbursed sensor would take share, as Abbott’s unit-volume leadership demonstrates. Brand is a tiebreaker, not a barrier.

What is the nature of competition? Three-dimensional: accuracy and wear duration (where Dexcom currently leads at the top of the range), price and payer economics (where Abbott leads), and ecosystem integration (where Dexcom leads on depth but has no exclusivity). Competition is increasingly won in the payer’s office rather than the physician’s — the firm with better formulary placement and lower net cost to the plan wins the expanding Type 2 population.

Customers’ switching costs? Real in the intensive-insulin tier, weak elsewhere. An AID user faces retraining, re-pairing, clinical revalidation and the loss of glucose history their clinician reads. Empirically, CONNECT showed 97% median wear over 26 weeks and Dexcom’s commercial registry shows above-80% utilisation in non-insulin patients — that is measurable stickiness. The limitation is that every AID partner (Tandem, Insulet, Beta Bionics) also supports Abbott, so the cost is friction, not lock-out, and it declines with each new Abbott integration. In the OTC/Stelo tier switching costs are essentially zero. (FACT/INTERPRETATION.)

Financial Condition & Balance Sheet

Assets not fully recognised on the balance sheet? Yes, and they are the important ones. Dexcom carries $24.1M of goodwill and $101.6M of net intangibles on $6,455.3M of total assets — essentially nothing. The genuinely valuable assets are all internally generated and therefore unrecognised: the FDA/CE clearance portfolio, the CONNECT and DIAMOND/MOBILE clinical evidence base, the AID integration relationships, the payer contracts covering the four largest PBMs, the manufacturing process knowledge that produced 400bp of scrap-driven margin, and the installed patient base itself. Book equity of $2,621M against a ~$32B market capitalisation is the measure of how much of this company’s value is off the balance sheet. (FACT/INTERPRETATION.)

Off-balance-sheet liabilities? Nothing unusual disclosed. Operating leases are capitalised ($88.0M of right-of-use assets against $20.7M short-term and $82.6M long-term lease liabilities). There are guarantee facilities related to certain international operations, partially collateralised and included in non-current other assets, which the company describes as not significant. The material contingent liabilities are legal — the securities class action, derivative actions and G6/G7 product suits — and are unaccrued and unquantified because the company states it cannot reasonably estimate the outcome. That is the real off-balance-sheet exposure. (FACT.)

How conservative is the accounting? Reasonably conservative, with two things to watch. Positives: revenue recognised on shipment with no long-dated contract accounting; no capitalised development costs of consequence; no goodwill to impair; SBC expensed and falling in absolute dollars ($39.0M in Q2 versus $45.3M); and management guides and reports both GAAP and non-GAAP with the reconciliation available. Watch items: (a) the non-GAAP/GAAP gap at the operating line was ~80bp in Q2 (25.1% versus 24.33%) — small and consistent, not abusive; (b) inventory rose 15.5% in the half against 12% revenue growth, which I read as the 15-day and Ireland build rather than channel stuffing, particularly because receivables rose only 3.4%, well below revenue. © Note that Q3 2025 GAAP net income ($283.8M) exceeded operating income ($242.5M) on an $81.8M non-operating gain, so trailing GAAP net income embeds a non-recurring benefit — a reason to anchor on operating income and ROIC rather than reported EPS. (FACT/INTERPRETATION.)

How CapEx-hungry is the business? Moderately, and currently less so. H1 2026 capital expenditure was $161.3M, down from $181.1M, roughly 6.5% of revenue annualised versus ~8% previously, funding Arizona, Malaysia and the Ireland facility starting commercial production later in 2026. Capex is presently running below depreciation (~$68M per quarter of D&A against ~$81M per quarter of capex — close, and below on a trailing comparison against the prior year’s rate), which flatters current free cash flow and cannot persist indefinitely in a company opening factories. This is not an asset-light software business; it is a manufacturer whose capital intensity is temporarily favourable.

Capital Allocation & Management

How much FCF does the business generate, how does management use it, what is the philosophy? H1 2026 free cash flow was $633.5M ($794.8M operating cash flow less $161.3M capex), against $305.7M in H1 2025 — a 107% increase on 12% revenue growth. Trailing free cash flow to the firm is roughly $1.42B. The philosophy is now formally stated for the first time: “at least 50% of cash generated from operations, net of planned capital expenditures, to share repurchases over the course of our long-range plan”, with the balance funding capacity expansion and technology-accelerating tuck-in acquisitions. No dividend. Given a 19.2% ROIC, retaining capital for reinvestment is the correct choice and the absence of a dividend is a feature. (FACT.)

Significant acquisitions recently? One, and it is small: Nutrisense, completed in Q2 2026 for $41.2M net of cash acquired — 0.13% of market capitalisation. Goodwill was unchanged at $24.1M while intangibles rose $70.8M → $101.6M, so consideration went almost entirely to identifiable technology. Nutrisense was a long-standing Dexcom partner; most of its revenue was already Dexcom Stelo pass-through, and management sized incremental non-CGM revenue at “the millions, you can count on one hand.” The rationale is CGM-guided nutrition insight and coaching technology to be integrated across Stelo and the G-series apps. Interpretation: the right kind of deal — a known partner, a technology asset, a trivial price, no goodwill. But it ends the clean “no acquisition history” characterisation and the Investor Day established a stated appetite for further technology tuck-ins, which is a new surface to monitor. (FACT/INTERPRETATION.)

Buying back shares? Yes, and this is the standout capital-allocation fact of the period. In May 2026 the board terminated the 2025 $750M program with $250M unused and authorised a $1.00B 2026 program running to June 30, 2027. In Q2 alone Dexcom repurchased 8.6 million shares for $600.0 million at an average of approximately $69.77 — the largest quarterly repurchase in its history, executed roughly 21% below the current price, into the year’s trough. Cumulative repurchases now approximate $3.1B since 2022. Critically, this reverses the earlier report’s principal criticism of pro-cyclical timing. (FACT.)

Issuing large amounts of new shares to insiders? No. Share-based compensation was $39.0M in Q2 2026 versus $45.3M a year earlier — falling in absolute dollars while revenue grew 13% — and $82.3M for the half, roughly 3.2% of revenue. Unrecognised RSU/PSU cost is $263.6M over a 2.2-year weighted-average period. The Q2 buyback alone was more than seven times the half-year SBC charge, so the share count is falling, not rising. The residual dilution risk is the $1.25B 2028 convertible notes, partially hedged by capped calls; diluted shares (390.1M) exceed basic (382.2M) by ~2.1%. (FACT.)

Compensation policy of directors/management? Say-on-pay passed at the May 27, 2026 annual meeting with 277,111,475 for against 36,664,959 opposed — 88.3% support of votes cast, with 640,924 abstentions and 22,107,994 broker non-votes. All twelve director nominees were elected; director Nicholas Augustinos drew the most opposition at 47,957,634 against (84.7% support). Interpretation: 88% is passing but below the ~95%+ typical of large caps with uncontroversial pay, consistent with residual shareholder scepticism about the growth-weighted incentive design that the earlier report criticised. The board did not fail any vote, but it was not endorsed enthusiastically either. (FACT/INTERPRETATION.)

Motivations of management? Mixed evidence, on balance improving. In favour: a published, dated, quantified five-year plan that management can be held to; the largest buyback in company history executed at the low rather than the high; a formal ≥50%-of-FCF return policy; and operational metrics (NPS, complaints, scrap) that management now volunteers as accountability measures. Against: insider ownership remains well under 1% of the company, and every Form 4 filed between June 1 and August 4, 2026 was a sale — former CEO and director Kevin Sayer sold 26,756 shares twice (July 6 at $72.00 and July 20 at $76.32), director Mark Foletta sold 4,000 shares across July 31 and August 3 at $80.00 and $85.32, and officers Michael Brown and Jon Coleman made comparable regular sales. All were pre-arranged Rule 10b5-1 transactions and therefore carry little signal — but the contrast with the incoming CEO’s ~$1M open-market purchase at the $55 lows is worth noting: there is no equivalent buying at $85. (FACT/INTERPRETATION.)

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? No. DexCom, Inc. is a Delaware corporation listed on the Nasdaq Global Select Market under DXCM, common stock, $0.001 par value. Ordinary Form 1099 treatment; no K-1.

Dividend policy? No dividend has ever been paid, and the amended credit agreement contains covenants limiting dividends and other restricted payments. Capital is returned exclusively through share repurchase, now formalised at a minimum of 50% of free cash flow. (FACT.)

How profitable is the business? See above — TTM operating margin 22.9%, net margin 20.1%, ROIC 19.2%, ROE 38.5%, with a trailing incremental operating margin of 67.6%, meaning roughly two-thirds of each incremental revenue dollar over the past year converted to operating profit. FY2026 guidance calls for a non-GAAP operating margin of 23.5–24% and an adjusted EBITDA margin of 31.5–32%, against a 2030 target of 29–30% and 36–37% respectively.

Is net income diverging from cash from operations? Yes, favourably. H1 2026 operating cash flow of $794.8M compares with GAAP net income of $448.6M (Q1 $199.5M plus Q2 $249.1M) — a conversion ratio of roughly 1.77x, driven by D&A of ~$135M, SBC of $82.3M and working-capital timing. Free cash flow of $633.5M still exceeds net income by ~41%. Divergence in this direction, sustained over multiple quarters, is a quality-of-earnings positive. The one caveat is that part of the gap comes from capex running below depreciation, which is a timing benefit rather than a structural one. (FACT/INTERPRETATION.)

Where does the valuation sit? At $84.75 (2026-08-07 close): EV ~$31.4B, EV/TTM revenue 6.3x, EV/TTM EBITDA 22.4x, EV/TTM EBIT 27.6x, trailing GAAP P/E 33.7x diluted, 33.0x FY2026 consensus EPS of $2.57 and 27.8x FY2027 consensus of $3.05. Against its own approximately ten-year history the composite valuation percentile is 9.8 (P/B 9.6th, P/S 11.6th), versus roughly the 3rd percentile two months ago. (FACT, with the P/E-percentile stale-EPS caveat noted in the memo.)

Risks & Downside

What factors would cause the stock to decline? In descending order of likelihood-weighted impact: (1) multiple compression, since at 27.6x trailing EBIT the 2030 plan is largely capitalised and delivery earns only an ordinary return; (2) an adverse or delayed CMS decision on non-insulin Type 2, expected before year-end 2026; (3) gross margin failing to resume expansion in 2027 after the guided Ireland-driven pause; (4) competitive bidding from 2028 taking more ASP than the long-range plan budgets; (5) an Abbott price escalation in Type 2 non-insulin and OTC; (6) escalation of the unresolved March-2025 FDA warning letter toward a consent decree; (7) a credibility relapse of the 2024 kind, which the market has demonstrated it will punish with 40% in a single session.

Risk of a catastrophic loss? Low. Dexcom holds net cash of roughly $549M, generates over $1.2B of annual free cash flow, has no near-term maturities before the May 2028 convertible, complies with its credit-agreement covenants, and sells a product with high clinical utility and demonstrated 97% wear compliance. The scenarios that would produce a catastrophic outcome are narrow: an FDA consent decree that restricted production for an extended period, or a technology discontinuity (credible non-invasive optical sensing) that made the entire consumable model obsolete. Both are low-probability; the second is very low and has been “five years away” for a decade.

Chance of a total loss? Negligible. A net-cash, cash-generative, category-leading manufacturer with a recurring installed base and no refinancing risk does not go to zero absent fraud. The realistic bad outcome is a de-rating to the mid-$50s on a growth and margin disappointment — roughly the November 2025 low — not an impairment of the enterprise.

Recent News & Events

Has the business environment changed recently? Yes, materially and mostly favourably, in five ways. (1) Commercial reimbursement is essentially won — all four largest US PBMs now cover Dexcom CGM for all people with diabetes, over 7 million Type 2 non-insulin lives, and the binding constraint has shifted to prescriber awareness, with ~9 million Americans covered but not using CGM. (2) A CMS national decision on non-insulin Type 2 is now dated, expected before year-end 2026 with coverage modelled from mid-2027 across a ~25 million population. (3) Clinical evidence strengthened decisively — the CONNECT randomised trial readout at ADA on June 6, 2026 showed a 0.9% A1C advantage versus routine care, 97% median wear, and the largest relative benefit in the GLP-1-only cohort, which directly contradicts the GLP-1 bear narrative. (4) Competitive bidding moved from open threat to budgeted event, with management stating the long-range plan assumes it begins in 2028. (5) The margin recovery was confirmed and extended — Q2 non-GAAP gross margin of 64.1%, up ~400bp year-over-year, with FY2026 guidance raised on every line. (FACT.)

Significant acquisitions? Nutrisense, June 2026, $41.2M net of cash — see above.

Change in accounting policies? None material. OECD Pillar Two global minimum tax rules, including transitional safe harbours, do not currently have a material impact. The California Franchise Tax Board commenced an audit of the 2022 and 2023 California income tax returns in June 2026; management does not expect significant adjustments. The FY2026 estimated effective tax rate is 21.8%, with an H1 actual of 23.6% elevated by share-based-compensation shortfalls and disallowed executive compensation. (FACT.)

Recent changes — new markets, facilities, management? Markets: Dexcom Flex launched in Germany for Type 2 basal and non-insulin; Health Canada cleared G7 15 Day on July 13, 2026, the first ex-US clearance for the 15-day platform; France and Canada were cited as the strongest international performers on recent access expansion. Facilities: the Ireland manufacturing facility begins commercial production later in 2026, with hiring, training and depreciation costs landing in Q3 ahead of output — the disclosed reason gross margin peaks in Q3 and steps down thereafter. Regulatory: Dexcom was named on July 22, 2026 the first company selected for the FDA’s TEMPO digital-health devices pilot. Management and board: Jake Leach is in his first full year as CEO with Jereme Sylvain as CFO; all twelve directors were re-elected at the May 27, 2026 annual meeting, including founder-era CEO Kevin Sayer, who now sits as a director and has been a regular 10b5-1 seller. Products: Stelo received the first-ever OTC CGM pediatric clearance on June 12, 2026 and a fully rebuilt app launched in late July that management describes as the platform for the future G-series app; G8 remains on track for 2027 submission and end-2027/early-2028 launch at half the size of G7, with ketones and potassium to follow glucose on a multi-analyte platform. (FACT.)

APPENDIX B — Source Appendix

DexCom, Inc. (NASDAQ: DXCM) · August 8, 2026

All sources accessed 2026-08-08 unless otherwise noted. Primary sources are listed first. This is a follow-up engagement; sources cited in the earlier report of 2026-06-08 that remain load-bearing are carried forward and marked accordingly.

A. Primary — SEC filings (CIK 0001093557)

Note on the May 14, 2026 8-K. The Investor Day 8-K could not be rendered into readable text through the extraction path used in this session (the SEC returned an empty body for the document). The two Investor Day facts that are load-bearing for this report — the $1.00 billion 2026 repurchase authorisation and the “at least 50% of cash generated from operations, net of planned capital expenditures, to share repurchases over the course of our long-range plan” policy — are independently confirmed in the Q2 2026 10-Q, which is primary. The 2030 margin and growth targets rest on secondary reporting of the Investor Day presentation (below) corroborated by management’s references to them on the Q2 earnings call, and are labelled as secondary-sourced facts in the memo’s Fact vs. Interpretation table.

B. Primary — Management commentary

C. Primary — Clinical, regulatory and product announcements

D. Quantitative data feeds

E. Secondary — news and market commentary

F. Prior published work on this name

G. Analytical frameworks

• Competition Demystified (Bruce Greenwald & Judd Kahn) — barriers to entry as the dominant consideration; the three genuine advantage types; the market-share-stability and ROIC tests. Applied in the Competitive Position section to name the moat as customer captivity plus economies of scale, and to run the share-stability test honestly against the 2024–25 share loss and subsequent recovery.
• Capital Returns (Edward Chancellor / Marathon Asset Management) — supply-side capital-cycle analysis; high returns attract capital and mean-revert. Applied in the Industry Dynamics section to revise the earlier mean-reversion call, and to locate the 2028 competitive-bidding event as the scheduled arrival of the cycle’s price-down.