Dexcom's Stelo cleared the OTC path in 2024. The hardware monopoly is the entire game.
CGM for non-diabetics is the most hardware-bound corner of the 2026 health-AI landscape. The category exploded in 2022 when Levels Health hit ~$80M raised by selling glucose-monitoring subscriptions to wellness consumers — but Levels never made a sensor. It rented Abbott Libre and Dexcom G7 hardware, paid a telehealth provider to write the necessary scripts, and built software on top. Then March 2024 changed the rulebook: FDA cleared Dexcom's Stelo as the first OTC continuous glucose monitor — no prescription needed, sold direct to consumers for ~$89/month. Abbott followed with Lingo and Libre Rio. Software-only entrants who lived off the prescription-CGM arbitrage suddenly faced a question: when Dexcom sells the same hardware directly with a clean app, what's your wedge? Levels has since pivoted aggressively toward an AI-coach layer. Nutrisense rebuilt around RD coaching. January AI doubled down on AI-driven food response prediction. The honest read for 2026: capital required is high (Series A minimum, $5M+); solo not viable; hardware partnership with Dexcom, Abbott, or one of the small Chinese/Korean OEMs is non-negotiable. The only paths that work are (a) a software/AI layer on top of OTC hardware with a clinical wedge Dexcom won't build, or (b) selling RD coaching that bundles CGM rental into the price. Pure data-tracking apps without a clinical or behavioral angle are dead on arrival.
March 2024 FDA clearance reset the category. No prescription, direct-to-consumer, designed for the non-diabetic wellness market. Hardware + companion app + cellular sync. Every software entrant now lives or dies by whether they have a Stelo data partnership.
Abbott's response. Lingo targets the performance wellness segment, Libre Rio targets prediabetes. Direct-to-consumer, smaller patch, longer wear. Competes head-on with Stelo on price and brand presence in Apple Health / Google Fit ecosystem.
The category-definer. Coined the "metabolic fitness" narrative, built a cult community among biohackers, athletes, and longevity nerds. Post-Stelo pivot: doubling down on AI-driven food/exercise/stress coaching layered on top of OTC sensors. Still the consumer brand to beat — but the hardware monopoly is gone.
Pivoted hard in 2024 from "data dashboard" to "RD-led coaching with CGM included." Closer to the Nourish playbook than the Levels playbook. The pivot validates that the durable moat is the human coach, not the hardware.
The technical wager: AI predicts your glucose response to a food without you actually wearing a sensor (after an initial 2-week calibration). If it works at scale it eliminates the recurring hardware cost. The cleanest "AI as the moat, hardware as the calibration data" play.
Finnish team, strong design and product velocity, much of the same playbook as Levels with a European twist (CE marking, EU privacy framework). Proves the model can survive outside the US Dexcom ecosystem.
Not a CGM but the closest adjacent — a different biomarker (breath CO2 → metabolic flexibility). Useful to study as a counter-example: hardware-first wellness gadgets can scale, but go public-via-SPAC paths almost always pressure margins.
Sells the same Dexcom hardware bundled with weight-loss coaching at $179/month. Differentiated on outcomes-driven copy ("lose weight with your glucose curve"). Cautionary watch: how do you defend "lose weight via CGM" against the GLP-1 wave?
Hardware data access at scale isn't a developer-portal API; it's a signed BAA + commercial agreement + minimum volume commitment, negotiated by ex-Dexcom/Abbott BD or someone with peer-level introductions. Without this, you're rebadging Stelo or scraping HealthKit — not building a real product.
PCOS, perimenopause, GLP-1 companion, postpartum, type-1 caregiver, athlete fueling, prediabetes reversal — narrow clinical contexts where CGM data has a known intervention path. Generic "see your spikes" is what Stelo's own app does for free.
If your software interprets CGM data and recommends action, you're likely SaMD (Software as a Medical Device) class I or II under FDA. Quality system regulation, design controls, 510(k) or De Novo pathways. If that sentence sounds foreign, find a regulatory advisor before writing code.
Dexcom spent ~$1B and 25+ years getting from concept to OTC. Abbott similar. The semiconductor + enzyme chemistry + miniaturized electronics + FDA pathway moat is real and not crossable on $5M of seed capital. The only sensor startups that survive are buying Chinese OEM hardware and licensing it — and even that path is brutally capital-intensive.
Dexcom's own app, Abbott's Lingo app, and Apple Health/Google Fit will normalize CGM data viewing as a free OS feature within 24 months. The moat is the intervention you build on top — clinical context, behavior change, coaching — not the dashboard.
As of late 2024, Stelo and Libre Rio are HSA/FSA-eligible under IRS guidance. This is the structural cashflow tailwind for the category — if you're not building HSA-attested receipts and FSA-store distribution into the GTM, you're leaving ~30% effective price advantage on the table.
Founders with healthtech ops + regulatory background + warm Dexcom/Abbott intro
RD/MD with audience or strong clinical credibility, mid-size capital
Solo or two-person team, no clinical claims, App Store distribution
Ex-Dexcom, ex-Abbott, ex-Verily, ex-Oura operators with hardware-partnership instincts and 510(k) experience. This is the highest-value background in this track — domain access compounds with every conversation.
Endocrinologists, OB-GYNs (PCOS / perimenopause), pediatric type-1 specialists, sports-medicine MDs, dietitians — the clinical wedge cannot be Googled into existence. If you treat patients in one of these niches, you're a candidate for the founder seat.
Relationships at large health systems, ACOs, or self-insured employers open a B2B2C path. CGM data + clinical wedge sold via existing payer channels is slower but stickier than the D2C path.
5 min · 12 questions · Free · Get your archetype + top 3 matching tracks
Take the quiz →