Anthropic Just Filed for a Trillion-Dollar IPO
On June 1, 2026, Anthropic, the company behind the Claude family of artificial intelligence (AI) models, confidentially filed for an initial public offering (IPO) that could value it in the trillion-dollar range [1]. The filing caps one of the fastest valuation climbs any company has ever recorded: $380 billion in February, then $965 billion in May after a $65 billion raise, surpassing rival OpenAI's last reported $852 billion [1]. Anthropic is one of three trillion-class AI debuts expected this year, alongside OpenAI, which filed its own S-1 a week later, and SpaceX [1] [5].
Numbers that large are easier to read as a climb, so here is every funding round from inception to the filing:
Date | Round | Raised | Valuation |
|---|---|---|---|
May 2021 | First funding round, the year seven OpenAI alumni founded the company [12] | $124 million | Estimated ~$500M–$700M |
April 2022 | Funding round, including $500 million from FTX [12] | $580 million | Estimated ~$4B |
September 2023 | Amazon strategic investment begins ($1.25 billion of a planned $4 billion, completed March 2024) [12] | $450 Milion | $4 billion |
October 2023 | Google invests $500 million and commits $1.5 billion more [12] | Up to $2 billion | $15 Billion |
November 2024 | Amazon doubles its total investment to $8 billion [12] | +$7.3 Billion | $18.4 Billion |
March 2025 | Series E, led by Lightspeed Venture Partners; Google adds another $1 billion [12] | $3.5 billion | $61.5 billion |
September 2025 | Series F, co-led by Iconiq Capital, Fidelity, and Lightspeed [12] | $13 billion | $183 billion |
December 2025 | Term sheet led by Coatue and GIC [12] | $10 billion | $350 billion |
February 2026 | Series G [1] [12] | $30 billion | $380 billion |
May 2026 | Series H, with Altimeter, Dragoneer, and Sequoia Capital [1] [12] | $65 billion | $965 billion |
June 1, 2026 | Confidential IPO filing [1] | n/a | Trillion-dollar range (potential) |
Worth noting these are all estimates and should be fairly close but may not be exact as all the numbers were not public
The first round was $124 million. Five years later the company raised more than 500 times that in a single round, and the last leg of the climb, $380 billion to $965 billion, took three months.
For medicine, the more interesting story is what the company shipped eight days after the filing.
On June 9, Anthropic launched Claude Fable 5, which anyone can use, and Claude Mythos 5, a restricted version whose protein design work Anthropic says ran about ten times faster than its own human experts [2]. That is the release worth understanding, because it touches three different layers of medicine at once: the everyday care AI is already part of, the practice tools a doctor can use today, and the drug pipeline that produces the medicines your doctor will prescribe in ten years.
AI Is Already in Everyday Medicine
The Mythos launch did not arrive in a vacuum. The same week, two unrelated announcements showed how fast AI is moving into routine care.
Google published a study in Nature describing a system that measures heart rate using nothing but a phone's front-facing camera [3]. In the seconds after you unlock the phone, an on-device neural network reads the changes in how light reflects off your skin as blood pulses through it. Built on more than 350,000 video clips from nearly 700 participants, the system estimates heart rate with a mean error under 10% against an electrocardiogram (ECG), and daily resting heart rate within 5 beats per minute of a wearable tracker [3]. Google flagged a real limitation: accuracy drops on darker skin tones [3]. Still, a vital sign that used to require a device now requires the phone already in your pocket.
The same week, Medicare Advantage insurer Essence Healthcare expanded its partnership with smart ring maker Oura into a clinical screening program for obstructive sleep apnea, a condition that affects 20% to 30% of adults over 65 but goes undiagnosed in roughly 80% of cases, according to Essence's chief medical officer, Saria Saccocio, M.D. [4]. With a member's consent, the ring's overnight breathing data flags people at risk, who are walked through STOP-BANG, an evidence-based sleep apnea screen, and the results go to the member's own physician to decide next steps [4]. Note who makes the final call: the doctor.
The biggest AI companies are building for that clinical layer directly. OpenAI's ChatGPT for Healthcare is deployed at systems including HCA Healthcare, Cedars-Sinai, Memorial Sloan Kettering, and UCSF, and its ChatGPT for Clinicians gives verified US doctors cited answers from medical sources at no cost [5]. Once Anthropic and OpenAI are public companies, the economics of all of this, what AI in medicine actually earns, becomes quarterly public record [1] [5].
The Claude Tools a Concierge Doctor Can Use Today
Mythos 5 is the research showpiece, but Anthropic's models have been doing quieter work inside medical practices all year, and most of it lands exactly where independent concierge and direct primary care (DPC) doctors feel the most pain: charts, paperwork, and insurance.
The clearest example is Elation Health, the electronic health record (EHR) built for primary care and used by more than 46,000 clinicians. In January, Elation embedded Claude into its EHR to power Clinical Insights, a feature that reads a patient's longitudinal chart, problem lists, medications, visit notes, labs, and vitals, and produces a concise pre-visit summary in which every line cites back to its source in the chart [6]. Elation reports clinicians get answers 61% faster at the point of care since the move to Claude, and use of the feature has tripled [6]. Co-founder and chief executive Kyna Fong, Ph.D., describes the problem it solves as clinicians being "data rich and insights poor" [6]. We covered Elation's broader push when it bought the voice-AI startup Aster to build a full AI front office.
Anthropic also sells into practices directly. Claude for Healthcare, launched in January, is a set of HIPAA-ready tools (HIPAA is the federal health-privacy law) for providers, payers, and health tech companies [7]. For a small practice, the useful parts are concrete: Claude can pull from the Centers for Medicare & Medicaid Services (CMS) Coverage Database to check coverage requirements and build stronger claims appeals, look up ICD-10 diagnosis and procedure codes for billing, verify providers against the National Provider Identifier registry, run a customizable prior-authorization review workflow, and triage patient portal messages so the urgent ones surface first [7]. The companies building practice tools run on the same models: Commure, whose ambient AI documents tens of millions of appointments, runs on Claude's model family [7], a thread we followed when Commure raised $70 million to automate medical billing with AI.
There is a patient side too. In the US, Claude Pro and Max subscribers can opt in to connect their own health records and lab results, with Apple Health and Android Health Connect integrations rolling out, so the assistant can explain test results in plain language and prepare questions for an appointment. Anthropic says the data is opt-in, user-controlled, and not used to train its models [7].
Here is why this matters most for concierge medicine. The single number that explains why care feels rushed is panel size: a traditional primary care doctor carries 2,000 to 2,500 patients, which is why you get 10 to 15 minutes. A concierge physician carries under 300, a DPC doctor up to 800, and visits run 30 to 60 minutes [11]. The Society of Actuaries' 2020 evaluation of DPC found patients visited the emergency room 40.51% less often than insurance-only patients [11]. Time and attention are the product. Every hour of chart review and prior-auth paperwork that Claude absorbs makes that product cheaper to deliver, which is why the EHR these practices run on adopted it first. In San Francisco, Anthropic's home city, practices like Golden Gate Concierge Medicine and the flat-fee clinic Castro Direct Primary Care compete on exactly that unhurried attention.
The 10x Claim: Drug Design at Machine Speed
Now the headline layer. Fable 5 and Mythos 5 are the same underlying model; the difference is the guardrails. In the public Fable 5, queries touching biology, chemistry, or cybersecurity fall back to a less capable model, Claude Opus 4.8, in fewer than 5% of sessions [2]. Mythos 5 has some of those safeguards lifted and is restricted to vetted partners through Project Glasswing, Anthropic's program with the US government, with a small trusted-access program for biology researchers planned [2]. The caution is not theater: in April, Treasury Secretary Scott Bessent and then-Federal Reserve Chair Jerome Powell warned major bank chief executives about the cybersecurity risks of the predecessor model [8].
What the unrestricted model did is the news. Anthropic's internal protein design experts report Mythos 5 accelerated aspects of the drug design process by around ten times, executing the full workflow a scientist normally would: choosing binding sites on a target protein, selecting and running protein design tools, and recovering from failures along the way [2]. In blinded comparisons, Anthropic's scientists preferred Mythos 5's molecular biology hypotheses about 80% of the time, and one hypothesis, a proposed mechanism for an E. coli protein, was independently corroborated by an outside lab [2]. Laid side by side, the reported results look like this:
What Mythos 5 did | Result | How it was measured |
|---|---|---|
Drug-design workflow speed | ~10x faster | Anthropic's internal protein design experts, with no human assistance on the task [2] |
Protein targets yielding strong drug candidates | 9 of 14 | Internal study spanning immune checkpoints, neurodegeneration, and muscle disease [2] |
Molecular biology hypotheses scientists preferred | ~80% of the time | Blinded head-to-head against Anthropic's earlier Opus-class models [2] |
Autonomous genomics research | Beat a model published in the journal Science at 1/100th the size | Over a week of largely autonomous work across 138 animal species [2] |
Safety fallback rate in the public Fable 5 | Fewer than 5% of sessions | Biology, chemistry, and cybersecurity queries routed to Claude Opus 4.8 [2] |
Every row carries the same asterisk: measured by Anthropic, on Anthropic's model.
Two caveats belong up front. These are company-reported results, not peer-reviewed studies, and the genomics work is not yet published [2]. And a candidate molecule is not a medicine: it still faces preclinical testing, clinical trials, and Food and Drug Administration (FDA) review, a process measured in years. The model also stays locked away from most working biologists for now, because the same skill that designs a gene therapy delivery virus could, in the wrong hands, design something dangerous [2] [8]. If the claims hold up, the realistic outcome is not faster cures next year. It is a much larger supply of credible drug candidates entering the pipeline, which over time means more shots on goal against diseases that have none today.
Who Built Anthropic
Anthropic was founded in 2021 by seven former OpenAI employees, led by Dario Amodei, the chief executive and previously OpenAI's vice president of research, and his sister Daniela Amodei, the president, who spent about three years at OpenAI and was earlier a founding recruiter at the payments company Stripe [9] [10]. The siblings left OpenAI over differences about the company's direction [10]. Five years later, their company is filing for what may be a trillion-dollar listing [1].
The Honest Read
A trillion-dollar IPO filing and a 10x drug-design claim landed eight days apart from the same company, and the temptation is to treat both as hype. The evidence cuts finer than that. The practice-layer tools are real and measurable today: cited chart summaries inside the EHR 46,000 clinicians use, prior-auth workflows, ambient documentation [6] [7]. The research-layer claims are promising and unproven, reported by the company itself and untested by peer review [2]. And none of it changes the part that was always the hard part. An AI can summarize your chart, flag your sleep apnea risk, and design a candidate molecule, but it is a model, not your doctor. The faster the science moves, the more valuable the physician with a small panel and a 45-minute appointment becomes, because someone still has to decide what all of it means for you.
Search concierge doctors near you on NextMD →
A Note From the Author
I am not a doctor. Nothing in this article should be considered medical advice. This piece is a plain-language explanation of an AI company's products and claims, and how they fit into the broader concierge and direct primary care landscape. Before making any change to your medical care, talk to a licensed physician who knows your medical history.
NextMD helps you find and compare concierge medicine and direct primary care practices across the United States. Browse practices by city, compare pricing, and find a doctor who has time for you at nextmd.ai/search.
Sources
Duffy, C., Eadicicco, L., & Gold, H. (2026, June 1). Anthropic files to go public in a potentially trillion-dollar debut. CNN Business. Read on CNN. Referenced for the confidential S-1, the trillion-dollar-range framing, and the $380B/$965B/$852B valuations.
Anthropic. (2026, June 9). Claude Fable 5 and Claude Mythos 5. Anthropic News. Read the announcement on Anthropic.com. Referenced for the launch, safeguards and fallback behavior, the ~10x drug-design acceleration, 9-of-14 protein targets, hypothesis preference, genomics results, and Project Glasswing access.
Hagen, J. (2026, June 8). Google unveils smartphone camera-based heart rate monitoring system. MobiHealthNews. Read on MobiHealthNews. Referenced for the Nature study, dataset size, accuracy figures, and skin-tone limitation.
Minemyer, P. (2026, June 8). Essence Healthcare, Oura expand partnership to sleep apnea detection. Fierce Healthcare. Read on Fierce Healthcare. Referenced for the screening program design, prevalence and underdiagnosis figures, and physician routing.
Bender, E. (2026, June 10). OpenAI, Anthropic take steps toward IPOs as healthcare AI footprints expand. MobiHealthNews. Read on MobiHealthNews. Referenced for OpenAI's filing a week after Anthropic's, ChatGPT for Healthcare deployments, and ChatGPT for Clinicians.
Landi, H. (2026, January 12). JPM26: Elation Health embeds Anthropic's Claude into EHR to power chart summaries. Fierce Healthcare. Read on Fierce Healthcare. Referenced for the 46,000 clinicians, 61% faster answers, tripled usage, and Kyna Fong quotes.
Anthropic. (2026, January 11). Advancing Claude in healthcare and the life sciences. Anthropic News. Read the announcement on Anthropic.com. Referenced for Claude for Healthcare's HIPAA-ready connectors (CMS Coverage Database, ICD-10, NPI registry), the prior-authorization skill, patient-message triage, Commure running on Claude, and the personal health-record integrations.
Hagen, J. (2026, June 10). Anthropic unveils Claude Fable 5 and Mythos 5 with safeguards. MobiHealthNews. Read on MobiHealthNews. Referenced for the Bessent/Powell warning to bank executives and the dual-use biosecurity rationale.
Forbes. (2026). Daniela Amodei profile. Forbes Profiles. Read on Forbes. Referenced for the 2021 founding with six other former OpenAI employees, Daniela's OpenAI and Stripe background, and the $965 billion May 2026 valuation.
Wikipedia contributors. (2026). Dario Amodei. Wikipedia. Read on Wikipedia. Referenced for Dario's role as OpenAI's vice president of research and the founders' departure over directional differences.
Busch, F., Grzeskowiak, D., & Huth, E. (2020). Direct Primary Care: Evaluating a New Model of Delivery and Financing. Society of Actuaries / Milliman. Read the SOA report (PDF). Referenced for the 40.51% emergency-room reduction and the DPC model evaluation; panel-size ranges per NextMD's canonical pricing and panel-size standards.
Wikipedia contributors. (2026). Anthropic. Wikipedia. Read on Wikipedia. Referenced for the funding-round history: the $124 million May 2021 round, the $580 million April 2022 round including FTX's $500 million, the Amazon ($4 billion plus $4 billion) and Google ($500 million plus $1.5 billion committed, plus $1 billion in 2025) strategic investments, Series E ($3.5 billion at $61.5 billion), Series F ($13 billion at $183 billion), the Coatue/GIC $10 billion term sheet at $350 billion, Series G ($30 billion at $380 billion), and the May 2026 $65 billion round at $965 billion with Altimeter, Dragoneer, and Sequoia.