Amwell’s Converge Turns Telehealth Into a $180 Billion Platform Bet

Converge Launch: Amwell Bets on Platform-as-a-Service

Amwell's co-founder Roy Schoenberg was named Amazon's health chief effective July 1, 2026, signaling a strategic shift toward platform-based healthcare infrastructure. Meanwhile, the global telemedicine market, valued at $85.5 billion in 2025, Research and Markets' data shows, is projected to reach $180 billion by 2031, Research and Markets' figures show, driven by multi-partner platform models rather than point-solution video visits.

Telehealth spent its first decade proving video visits could work. Its next decade will be defined by who owns the operating system.

The shift mirrors what happened in cloud computing fifteen years ago. Amazon didn't win by selling better servers; it won by letting developers rent compute, storage, and databases as composable services. Amwell is betting healthcare will follow the same pattern — that hospitals and health plans want to assemble virtual care experiences from best-of-breed components rather than buy monolithic suites. Amwell provides the rails: scheduling, eligibility verification, clinical documentation, prescribing workflows, and the HIPAA-compliant connective tissue that makes those pieces interoperate in production.

DarioHealth, a digital therapeutics company focused on cardiometabolic conditions, used Amwell's channel partnership to reach a major Arizona insurer's administrative-services-only population, accessing hundreds of thousands of covered lives through a single integration. The channel strategy lowers customer acquisition costs and accelerates sales cycles across employer populations; as of July 2026, Dario reports access to over 116 million covered lives through its expanding channel network, much of it routed through Amwell's platform.

This is app-store logic applied to regulated clinical delivery. Instead of each digital health vendor negotiating separate contracts, security reviews, and EHR integrations with every health system, they integrate once with Amwell's platform. Health systems gain a curated catalog of vetted digital programs they can activate for specific patient populations. Insurers get a configurable network of virtual services they can bundle into benefit designs. Amwell takes a platform fee and gains the data exhaust that comes from orchestrating the transactions.

The market context underscores the stakes. The global telemedicine and digital health market reached $85.5 billion in 2025 and could hit $180 billion by 2031, growing at roughly one in eight annually, Research and Markets found. North America holds more than a third. But the competitive dynamic has shifted: participants now compete on their ability to orchestrate the full continuum of care, spanning virtual diagnosis, e-prescriptions, remote patient monitoring, chronic care management, patient engagement, and outcomes analytics, within secure and interoperable digital environments. Point-solution video visits are table stakes. The margin sits in the orchestration layer.

Schoenberg's appointment at Amazon, replacing Neil Lindsay, signals how seriously the tech giants take the platform play. Amazon's own health ambitions (PillPack, One Medical, a shuttered telehealth service) have cycled through build, buy, and retreat. Tapping Schoenberg suggests the next phase isn't about owning clinics or pharmacies. It's about owning the rails.

Converge is still early. The partner roster beyond Cleveland Clinic has not been publicly detailed at scale. Reimbursement parity for multi-modal virtual care remains uneven across states and payers. And the technical complexity of real-time clinical data exchange across organizational boundaries (FHIR APIs, consent management, audit trails) is non-trivial. But the architectural bet is clear: telehealth is becoming a multi-sided platform business. The companies that build the operating system will set the terms for everyone else.

Teladoc's Struggles Show a Maturing Market

Teladoc's first quarter of 2026 laid the numbers bare. Revenue came in at $614 million, down 2.4% from $629 million a year earlier. The U.S. segment, still the bulk of the business, contracted 6.3% to $492 million. International revenue grew 17.3% to $122 million, not enough to offset the domestic slide. Adjusted EBITDA flatlined at $58.2 million, essentially unchanged from $58.1 million. The company still posted a net loss of 36 cents per share, even after restructuring costs. Operating cash flow: $10 million.

The BetterHelp unit tells the sharper story. Revenue fell to $218 million from $240 million in Q1 2025. Paying users dropped from 397,000 to 361,000. EBITDA collapsed to $2 million, down from $18 million just one quarter prior. Customer acquisition costs have risen as social-media privacy changes and new competitors bid up ad prices. The mental-health app model carries high natural churn; users don't stay forever. Teladoc is on a treadmill buying replacements.

Management knows it. On the earnings call, executives framed the response in three moves: tighten operational costs, expand internationally, and upgrade domestic users into chronic-care programs that stick. Chronic-care enrollment did grow, from 1.15 million to 1.20 million. U.S. integrated-care members held roughly flat at 101.2 million. The company projects full-year 2026 revenue of $2.48–2.58 billion, essentially zero top-line growth, with adjusted EBITDA of $267–367 million, a 5–9% improvement. They expect U.S. integrated-care members to hover between 97–100 million all year. The message: the hypergrowth era is over.

The broader pattern is clear: Teladoc built a generalist "everything to everyone" telehealth empire through billion-dollar deals (Livongo the largest) and now faces the integration bill. Analysts openly speculate a spin-out of the mental-health division could follow, leaving a specialized chronic-care platform.

That trajectory mirrors the market's consolidation. The spoils are concentrating in players with scale, clinical relationships, and regulatory infrastructure — Teladoc, Amwell, Philips, Oracle Health (Cerner), Siemens Healthineers. The pandemic forced every provider into telemedicine overnight; that mandate created a surge, not a sustainable model. As one analyst put it, high cultural demand doesn't equal a viable business if the mechanics of acquiring customers are broken.

Amwell's Converge platform bets on a different mechanic: a multi-partner marketplace where health systems and tech providers plug in their own digital offerings. Teladoc is betting on vertical integration — owning the clinicians, the chronic-care programs, the international footprint. Both are platform plays. Both are fighting the same physics: rising acquisition costs, payer scrutiny, regulatory weight. The difference is architectural. Teladoc owns the supply side; Amwell is building the rails for others to bring supply.

The consolidation wave is visible in the headcount data: Amwell laid off roughly 50 employees in March 2023 and another round in January 2024, a 10% headcount reduction since end of 2023. The market is sorting into platform tiers. The winners will be the ones who can operate multi-sided, regulated marketplaces at scale — not just run video visits.

Google Cloud Enters, Changing Health Tech Infrastructure

Google Cloud's partnership with Augmedix signals that hyperscalers now treat healthcare platforms as a distinct infrastructure layer — one that demands medically tuned AI, not generic cloud services. As of December 2023, the ambient documentation company piloted Med-PaLM 2 and moved MedLM, Google's healthcare-specific large language model, into production on Vertex AI. Augmedix's training data spans 70,000 notes per week across more than 30 specialties. HCA Healthcare had already deployed Augmedix products in four hospitals.

Google Cloud CEO Thomas Kurian framed the bet directly: generative AI can automate one of the most time-intensive and tedious tasks in healthcare, capturing accurate medical notes of doctor-patient interactions. Augmedix founder and chief strategy officer Ian Shakil added that healthcare-grade AI requires a more tailored and precise approach than general-purpose LLMs can provide. MedLM, now generally available to allowlisted customers, is built for that precision. It powers Augmedix Go, a clinician-controlled mobile app that produces a fully automated draft note after each visit, with expansion across all documentation products underway.

This matters for platform engineering because platforms like Converge are designed to host exactly these kinds of partner services (clinical programs, AI models, and others) on a single runtime. The architecture shifts from "video visit plus EHR integration" to "orchestrated marketplace of regulated microservices." Each partner brings its own data contracts, compliance boundaries, and model-update cadences. The platform must route requests, enforce consent, log audit trails, and version models without downtime. That is distributed systems work at a level most telehealth vendors have never needed.

Amazon's trajectory reinforces the pattern. After acquiring PillPack for roughly $750 million in 2018, CNBC's data shows, and One Medical for $3.9 billion in 2023, CNBC reported, Amazon launched and then shuttered its own telehealth service and Halo wearables line. On July 1, 2026, Neil Lindsay steps down as Amazon's top health executive; his replacement is Dr. Roy Schoenberg, Amwell's cofounder. Amazon also launched an AI health tool that analyzes medical records, books appointments, and answers queries. The message: even a company with Amazon's capital and distribution concludes that healthcare platforms require specialized clinical-technical leadership and partner-ready infrastructure — not just owned-and-operated services.

Cloud-based solutions dominate for flexibility and cost efficiency. Key players (Teladoc Health, Philips Healthcare, Oracle Health (Cerner), Siemens Healthineers) are embedding AI-driven clinical decision support, population health analytics, and interoperable EHR connectivity into their platforms. They are not buying point solutions; they are building partner ecosystems.

For engineers, the hiring bar moves accordingly. Experience with Vertex AI, MedLM, or equivalent medically tuned model stacks becomes a differentiator. So does fluency in FHIR-based data exchange, consent-management frameworks, and the operational discipline of running multi-tenant services under HIPAA and FDA scrutiny.

Engineering the Multi-Partner, AI-Powered Platform

Amwell's Converge platform does not just host video visits. It hosts other companies' clinical products — Cleveland Clinic's second-opinion service, digital therapeutics from partners like DarioHealth, and whatever partner joins next. That shift from single-vendor tool to multi-sided marketplace rewrites the engineering job description. The platform must route a patient's request to the right partner, pull real-time clinical data from an EHR, invoke an AI model that may live in another cloud, return a structured response, and log every step for audit. All in seconds. All under HIPAA.

Stanford Health Care's ChatEHR platform shows what this looks like in production. Their team tried the obvious path first: combine the EHR's nightly reporting database with HL7v2 real-time feeds. Reconciling those two sources "proved complex to maintain," the Stanford team wrote. They landed on Fast Healthcare Interoperability Resources (FHIR) instead. A serverless function now fetches and organizes clinical information using FHIR, with intelligent caching for frequent patterns and parallel processing that breaks complex queries into concurrent operations. The result: a system that processes millions of clinical data points while maintaining consistent performance.

The integration layer is where most platform efforts stall. Stanford built an enterprise integration service that manages secure connections between the ChatEHR platform, the EHR, and other IT systems — authentication, rate limiting, comprehensive logging, process automation, and scheduling. That service "provides a secure and reliable connection with the EHR" and creates the business logic for applications built on top. Amwell's Converge faces the same problem at scale: every new partner brings its own API contracts, data schemas, authentication flows, and compliance paperwork. Stanford's answer (a standardized integration layer that vendors plug into) eliminates costly, custom configurations for each new partner. The platform's existing standard integration, LLM, and evaluation capabilities become the onboarding pipeline.

Then there is the AI orchestration problem. Stanford deployed a self-hosted gateway for all large language model interactions, a single, secure access point that centralizes authorization, logging, and monitoring across model providers. An LLM router selects the correct model for each query type and routes it to a self-hosted server that standardizes all calls into a common format. A function server provides task-specific endpoints: specialized chat completion endpoints that combine LLM capabilities with clinical data access, transforming generic AI into healthcare-specific functions. This is not prompt engineering. It is platform engineering with models as runtime dependencies.

Deloitte's 2025 survey of 100 U.S. health care technology executives confirms the direction. Sixty-one percent say they are already building or implementing agentic AI initiatives or have secured budgets; 85 percent plan to increase investment over the next two to three years. Early adopters (mostly organizations above $5 billion in revenue) prioritize multi-agent solutions (82 percent). Watchers prefer point solutions (92 percent). The gap matters. Agentic AI can plan and sequence tasks, adapt to conditions, and coordinate with people and platforms to help deliver outcomes across clinical, administrative, and financial domains. That coordination requires engineers who understand distributed systems, event-driven architectures, and the semantics of clinical workflows — not just model serving.

The skill set is specific. FHIR fluency is table stakes. So is designing for eventual consistency across partner APIs that you do not control. Engineers must build observability into every hop (request tracing, latency budgets, error budgets) because a failed prior-authorization call or a dropped lab result is not a bug; it is a compliance incident. They must treat model outputs as untrusted inputs: validate, sandbox, audit. And they must do it in environments where technical talent limitations remain a major operational barrier cited by 60 percent of leaders, per Deloitte.

The marketplace model compounds everything. A platform that onboards Cleveland Clinic today and a radiology AI vendor tomorrow needs a partner SDK, a sandbox environment, automated contract testing, and a certification pipeline that verifies clinical safety before go-live. Stanford is building exactly that: a fifth capability domain focused on responsible evaluation, extending the platform from implementation to continuous learning and oversight, using the MedHELM framework. The engineers who can ship that pipeline (and keep it running as the partner count grows from three to thirty) are the ones the next wave of health tech platforms will fight for.

Regulation Makes Platform Engineering a Distinct Discipline

Amwell's Converge platform hosts Cleveland Clinic's virtual second-opinion service, the aforementioned digital therapeutics, and other clinical programs — each carrying its own regulatory footprint. That multiplicity transforms compliance from a checklist into an architecture problem. A point-solution telehealth vendor secures one data flow. A platform marketplace must secure every partner's data flow, prove each integration meets HIPAA's Security Rule, and demonstrate to the FDA that the platform itself does not become an unregulated medical device when a third-party algorithm runs on it.

The Alexa case illustrates the gap. When Amazon announced HIPAA compliance for its Alexa Skills Kit in April 2019, it limited access to six healthcare companies (pharmacy benefit managers and hospitals), each bound by a Business Associate Agreement (BAA). Drexel law professor Robert Field described the arrangement on the Knowledge@Wharton podcast: "This is kind of turning the notion of HIPAA privacy on its head. It's data coming in through the business associate." BAAs were designed for back-office processors, not consumer-facing voice interfaces that might broadcast protected health information to a room. The same inversion hits a multi-tenant platform: every partner onboarding becomes a BAA negotiation, every API contract a compliance artifact, every log retention policy a legal decision.

FDA oversight compounds the challenge. The agency's Software as a Medical Device (SaMD) framework and its growing focus on Digital Therapeutics (DTx) mean that a platform hosting clinical decision support or disease-management apps may itself fall under regulatory scrutiny. The market report identifies the rise of DTx and SaMD as a key trend, noting that established players use compliance-ready platforms to support large provider networks. For Converge, that means the platform layer must enforce boundaries: isolating a partner's DTx module so its clinical claims do not bleed into the core video stack, maintaining audit trails that satisfy 21 CFR Part 11, and supporting post-market surveillance data flows without violating patient consent.

State privacy laws add a third dimension. California's Consumer Privacy Act, Virginia's CDPA, Colorado's Privacy Act, and HIPAA's preemption patchwork require platforms to route data by jurisdiction, honor deletion requests across partner boundaries, and maintain consent ledgers that survive partner churn. The report lists data privacy, cybersecurity risks, and regulatory complexity as a top market restraint. Quinnipiac University's cybersecurity program, built with Clearwater's IRM|Pro platform, trains engineers specifically on healthcare enterprise cyber risk management — a curriculum that treats HIPAA Security Rule risk analysis, NIST 800-53 control mapping, and incident response under breach-notification timelines as core engineering competencies, not afterthoughts.

This regulatory stack creates roles that do not exist in consumer SaaS. Amwell's own job board lists a Manager, Government Compliance & Authorization (salary band $126k–$154k) and a Sr. Counsel, Government & Commercial ($170k–$208k): positions that blend cloud authorization, HIPAA attestation, and partner BAA management. The Technical Advisor role ($155k–$200k) sits at the intersection of platform architecture and regulatory evidence generation. Engineers who can design a FHIR-native API gateway that enforces consent directives at the field level, implement immutable audit logs that satisfy both HIPAA and FDA 21 CFR Part 11, and automate BAA clause extraction for partner onboarding are not "backend engineers with healthcare exposure." They are a distinct discipline: regulated-platform engineers.

Companies building multi-sided health marketplaces (Amwell, Teladoc, Oracle Health, Siemens Healthineers) are not recruiting for feature velocity alone. They are recruiting for evidence velocity: the ability to ship code that arrives pre-documented for auditors, pre-partitioned for partner isolation, and pre-instrumented for regulatory reporting. That skill set commands a premium because it cannot be learned from a bootcamp. It requires production scars from actual submissions, actual breach responses, actual FDA Q-Submissions. The platform era of telehealth will be built by engineers who treat compliance as a runtime constraint, not a launch gate.

What Roles Platforms Now Prioritize

The shift from video-visit tooling to multi-partner platform marketplaces is rewriting health tech job descriptions. Amwell's own board shows the transition in real time: two roles posted in the past week alone, spanning senior counsel for government and commercial contracts at $170k–$208k, a technical advisor at $154k–$200k, a senior product marketing manager at $136k–$166k, a government compliance and authorization manager at $126k–$154k, a lead FP&A analyst at $117k–$143k, and an implementation manager at $95k–$116k. The board's salary band clusters around a $166k median across seven salaried roles — compensation that reflects the premium for engineers who can operate inside regulated, multi-sided environments.

Broader tech is shedding headcount (more than 150,000 cuts across 549 companies in 2025, per Layoffs.fyi), yet health tech platforms are hiring for a different profile. Hims & Hers trimmed 68 roles (roughly 4% of staff) in late 2025. Nautilus cut 25 (16%). HerMD shut down entirely. Inside Inbound Health, a hospital-at-home startup that had raised over $50 million, closed its doors December 1. The pattern: point-solution telehealth vendors are consolidating or exiting, while platform-layer companies absorb the talent that understands API orchestration, partner onboarding pipelines, and real-time clinical data flows across organizational boundaries.

Deloitte's 2025 engineering outlook flags the same dynamic: migration of engineering talent to technology firms is intensifying competition for skilled workers, and advanced digital tools are increasing demand for data scientists, digital engineers, and specialists capable of managing AI-driven insights. In health tech, that demand carries a regulatory overlay. Illinois and California now prohibit AI from replacing mental health professionals; Texas bars algorithms from making adverse benefit determinations; Colorado's AI Act requires impact assessments before deploying systems that automate core job functions. Engineers who can build guardrails into the platform layer (audit trails, human-in-the-loop checkpoints, explainable model outputs) are becoming a distinct hiring category.

The roles clustering at the top of platform org charts reflect this stack:

• Platform/integration engineers who design multi-tenant APIs, versioned contracts, and sandbox environments for partner onboarding
• Regulatory/compliance engineers who translate HIPAA, FDA SaMD, and state telehealth statutes into automated policy enforcement
• Clinical data engineers who normalize FHIR resources, HL7 feeds, and partner-specific schemas into a unified event stream
• AI/ML platform engineers who deploy inference services with model cards, bias monitoring, and clinician review workflows baked in
• Technical product managers who coordinate roadmap dependencies across external partners and internal squads

Experience requirements skew senior. Amwell's technical advisor role sits at the top of its band; the compliance manager role demands fluency with cloud authorization frameworks and HITRUST. The median $166k on Amwell's board aligns with the biomedical engineering median of $108,060 reported by UND — but the platform premium pushes specialized roles 50% higher.

Hiring velocity will track platform adoption. Every new partner added to Converge (or to whatever platform Teladoc builds) creates a fresh integration surface, a new compliance matrix, and a new set of SLAs that must be monitored, versioned, and tested. The next partner onboarding won't be announced in a press release. It'll ship in a FHIR payload routed through a sandbox that passed automated contract testing at 2 a.m. — written by an engineer who treats HIPAA as a runtime constraint.

Working in frontier tech? Zero G Talent tracks the openings: see every open Amwell role, browse frontier tech jobs, the companies hiring, and the people building the field.