TrendingMay 29, 20268 min read·ByAyush Chaturvedi· Independent Entrepreneur

Anthropic Just Hit a $965B Valuation and Passed OpenAI. Here’s What It Means for Founders Building on Claude.

On May 28, 2026, Anthropic raised $65B at a $965B valuation, passing OpenAI as the most valuable AI startup. Here’s the playbook for founders building on Claude.

Key Takeaways

  • On May 28, 2026, Anthropic raised $65B at a $965B post-money valuation — nearly tripling its $380B value from February and passing OpenAI ($852B) to become the world’s most valuable AI startup.
  • The revenue is real, not circular: run-rate hit ~$47B (vs OpenAI’s ~$25B), up 50% in roughly seven weeks and driven by enterprise. Claude Code alone went from $0 to $2.5B ARR in nine months.
  • This is likely Anthropic’s last private round before an IPO. For founders, it de-risks the single biggest decision in an AI product: which platform to build on.
  • The play isn’t to compete with Anthropic — it’s to build on the open standards it controls (MCP, Agent Skills), ship into its zero-rev-share marketplace, and keep your moat in data and UX, not the model.

On May 28, 2026, Anthropic closed a $65 billion Series H at a $965 billion valuation — and quietly passed OpenAI to become the most valuable AI startup on earth. The number is staggering, but the headline isn’t the point. For the millions of founders building on Claude, the single most expensive decision in any AI product — which platform to bet on — just got a lot less risky. Here’s what actually changed, and the playbook to act on it.

What Anthropic Actually Did

The Series H raised $65 billion at a $965 billion post-money valuation — nearly triple the $380 billion Anthropic was worth in February, when it closed a $30 billion Series G. The round was co-led by Altimeter, Dragoneer, Greenoaks, Sequoia, Coatue, and Capital Group, with Baillie Gifford, Blackstone, Brookfield, Fidelity, and DST piling in. Tellingly, memory and chip makers Samsung, SK Hynix, and Micron joined too — the people who supply the compute now have equity in the demand.

TechCrunch reports this is likely Anthropic’s last private round before an IPO, and Fortune notes it would be among the first companies in history to approach a $1 trillion valuation while still private. The capital is earmarked for three things: research, compute to meet Claude demand, and the products and partnerships customers run on. On compute alone, Anthropic has committed north of $100 billion to AWS over a decade, $30 billion to Azure, and multi-gigawatt TPU capacity with Google and Broadcom.

The part that should get a founder’s attention isn’t the valuation — it’s the revenue underneath it.

The Raise by the Numbers

$965B
Post-money valuation
Up from $380B in February
$65B
Series H raise
Its largest private round yet
~$47B
Run-rate revenue
Up ~50% in seven weeks
$2.5B
Claude Code ARR
From $0 in May 2025
1,000+
Customers over $1M/yr
Doubled in under two months
#1
Most valuable AI startup
Passed OpenAI ($852B)

Sources: TechCrunch, CNBC, Fortune, Simon Willison (May 28–29, 2026).

Why This Matters: The Platform Bet Just Got De-Risked

Every AI product makes one bet it can’t easily reverse: which lab’s platform to build on. Pick wrong and you’re re-plumbing your stack when the vendor raises prices, deprecates a model, or runs out of road. Until this week, that bet carried real uncertainty. A $965 billion valuation backed by ~$47 billion of run-rate revenue — growing 50% in seven weeks — turns Claude into the safest infrastructure bet in AI.

And this is enterprise revenue, not a vanity metric. Anthropic crossed ~$47B run-rate versus OpenAI’s ~$25B, with over 300,000 business customers and 1,000+ accounts spending more than $1M a year — a number that doubled in under two months. Claude Code, the coding agent most of this audience already uses, went from $0 to $2.5B ARR in nine months, with enterprise now over half its revenue.

MetricAnthropicOpenAI
Post-money valuation$965B (May 28, 2026)$852B (Mar 2026)
Run-rate revenue~$47B~$25B
Revenue ~12 months ago~$10B~$13B
Latest round$65B Series H— (last raised Mar)
Most valuable AI startupYesNo (as of May 2026)

Sources: CNBC, Al Jazeera, Neowin (May 28–29, 2026). Figures are run-rate estimates.

Stay Ahead of the Trends

Get insights like this before they’re everywhere. Weekly breakdowns of the shifts that actually matter for indie hackers and SaaS founders. No fluff.

The Deeper Analysis: Three Things Most Coverage Is Missing

One: this revenue is real, which is the whole point. The loudest critique of the AI boom is circular financing — Nvidia funds the labs, the labs buy Nvidia, hyperscalers book it all as revenue. Analysts note ~$660–690B of 2026 hyperscaler capex against only ~$51B of direct AI revenue. Anthropic is the counter-example: ~$47B of run-rate revenue coming from outside customers — Netflix, Spotify, KPMG, L’Oréal, Salesforce — paying for Claude because it does work. When you build on Claude, you’re building on demand that exists, not demand that’s being manufactured between balance sheets.

Two: the ecosystem moves are the real tell. A lab confident in its position doesn’t hoard the platform — it opens it. Anthropic turned MCP and Agent Skills into open standards now governed by a Linux Foundation body with ~150 members, runs a zero-revenue-share marketplace, renamed the Claude Code SDK to the Agent SDK, and shipped native Claude into Xcode. That’s a company trying to become the default substrate other people build on. The valuation funds the compute that keeps that substrate fast and cheap.

Three: the IPO changes the incentives in your favor. A company marching toward a public listing needs durable, diversified, third-party revenue — exactly the kind that flows through a healthy developer ecosystem. Expect more credits, more distribution surface, and more reasons to ship on Claude over the next few quarters, because your success is now part of Anthropic’s S-1 story.

What You Can Do About It: The 5-Play Founder Playbook

You can’t out-raise Anthropic and you don’t need to. The opportunity is to build on the platform it’s pouring $65B into — capturing the distribution, the standards, and the subsidized inference, while keeping your moat somewhere the model can’t reach. Each play below is shippable from a laptop with a coding agent and a paid Claude plan.

1. Standardize on the Open Layers, Not the Model

The durable bet isn’t “Claude Opus 4.8 forever” — it’s the protocols Anthropic turned into industry standards. MCP (tool calls) and Agent Skills (packaged workflows) are now open specs governed by the Linux Foundation’s Agentic AI Foundation, adopted by OpenAI, Google, Microsoft, AWS, and Block. Build on those and you ride Anthropic’s momentum without locking yourself to one vendor’s pricing.

What to actually build

  • Make MCP your integration layer. Wire your product’s tools and data through MCP servers, not bespoke per-model glue. MCP has crossed ~97M monthly SDK downloads and every major lab speaks it — you write the connector once, it works everywhere.
  • Package your logic as an Agent Skill. A Skill is a folder of instructions plus scripts. The same Skill runs on Claude today and ports to other Skills-compatible runtimes tomorrow. Your IP lives in the Skill, not in a prompt you’ll rewrite every model release.
  • Treat the model as swappable. Keep an eval suite and an abstraction seam so you can route to the cheapest model that passes your bar. Building on the standard is the hedge; betting on one checkpoint is the risk.

2. Ship Into the Claude Ecosystem Directory

Anthropic’s marketplace takes zero revenue share from third-party developers — it monetizes via API usage, you keep 100% of your Skill, connector, or app revenue. With 300,000+ business customers already inside, a listing is distribution you don’t have to buy.

What to actually build

  • Build the missing connector for a vertical. The launch directory leans on partners like Atlassian, Canva, Cloudflare, Figma, Notion, Ramp, and Sentry. The long tail of vertical tools — ServiceTitan, Jobber, Procore, Mindbody — is wide open. First mover in a category becomes the canonical option.
  • Charge per outcome, not per seat. Usage-based pricing ($0.01–0.05 per invocation, or per cleared task) scales as a customer’s Claude rollout deepens. That’s the Twilio shape — and it compounds with the platform instead of capping at headcount.
  • Win the listing, not the ad auction. Distribution inside the directory is the listing itself. Your acquisition cost is shipping something genuinely useful and naming it for the workflow people search.

3. Let the Agent SDK Subsidize Your Margins

The renamed Claude Agent SDK (formerly the Claude Code SDK) gives you autonomous agents — subagents, background tasks, tool use, sandboxes — as a primitive. Anthropic is pushing inference credits to paid plans to seed the ecosystem. For a bootstrapper, that’s near-zero COGS for your first cohort of users.

What to actually build

  • Prototype on the Agent SDK, not raw API loops. You inherit the orchestration (planning, tool calls, retries) that you’d otherwise hand-roll. Time-to-MVP drops from weeks to days, and the same harness powers Xcode’s native Claude integration — proof it’s production-grade.
  • Use self-hosted sandboxes and MCP tunnels for sensitive data. Keep customer files and services in your own infra while the agent loop runs on Anthropic’s. That removes the single biggest objection from regulated SMB and enterprise buyers.
  • Apply credits to your burn. Free or discounted inference on paid plans means your bootstrapped MVP can serve 10–50 customers before token cost matters. Pocket the difference as runway.

4. Sell Picks-and-Shovels to the $1M+ Whales

Over 1,000 customers now spend more than $1M a year on Claude — that number doubled in under two months. They’ve adopted the platform; what they lack is the operational layer around it. You don’t have to find the buyers — Anthropic already did.

What to actually build

  • Build eval, observability, and governance. Teams running Claude in legal, finance, and accounting need to prove outputs are correct and auditable. Regression testing for prompts and Skills, cost dashboards, and approval logs are unglamorous and very sellable.
  • Productize the vertical layer on top. Skills + connectors give you ~70% of a vertical SaaS for free. Wrap the missing 30% — vertical UI, multi-user roles, industry reporting — and sell the bundle. Customers renew for the workflow, not the model underneath.
  • Become the “Claude ops” retainer. Eval new Skills, swap models as pricing shifts, monitor failures. Stack a dozen mid-market accounts at $1.5–3K/mo and you have a real business riding someone else’s sales team.

5. Keep the Moat in Your Data + UX — and Hedge the Bubble

Anthropic’s revenue is real external enterprise demand, which sets it apart from the circular Nvidia–hyperscaler–lab deals analysts are warning about. But the broader AI capex math is stretched (roughly $660–690B in 2026 hyperscaler spend against ~$51B of direct AI revenue). Build like the platform will keep winning — and like its pricing could still move against you.

What to actually build

  • Own proprietary data and distribution. The model is a commodity input; your customers’ data, your vertical workflow, and your audience are not. That’s what survives a model swap, a price hike, or a market correction.
  • Price with margin headroom. If your unit economics only work at today’s token prices, one pricing change breaks you. Charge for outcomes with enough spread to absorb a 2–3x input-cost move.
  • Stay portable. Building on open standards (Play 1) is also your bubble hedge — if any single lab stumbles, your MCP servers and Skills point at the next one. Bet on the ecosystem, not on any one company being immortal.

Looking Ahead: Three Things to Watch

Watch the IPO timeline. If this really is the last private round, an S-1 could land within 12–18 months. A public Anthropic means quarterly disclosure of exactly how its developer ecosystem is performing — the clearest signal yet of where to build. The pre-IPO window is also when platforms are most generous to the builders who make their numbers look good.

Watch OpenAI’s counter. Being leapfrogged in both revenue and valuation will provoke a response — likely aggressive pricing, a bigger developer push, or its own ecosystem land-grab. Healthy for builders: two trillion-dollar-class labs competing for your loyalty drives down your input costs. Stay portable so you can play them off each other.

Watch the bubble math. The capex-vs-revenue gap across the industry is real, and a credit tightening could hit valuations hard. Anthropic’s genuine revenue insulates it better than most — but build with margin headroom and an exit seam anyway. Riding the winner and hedging the downside are not contradictory.

Tools to Find Your Build-on-Claude Wedge

Pressure-test the vertical, generate the idea, and spec the Skill before you write a line of code.

Related reading: Anthropic’s Claude for Small Business Playbook and Stripe’s Agentic Commerce OS — both pair well with the build-on-the-platform thesis here.

The Bottom Line

  • Anthropic is now the most valuable AI startup. $65B Series H, $965B valuation, passing OpenAI in both valuation and revenue — likely its last round before an IPO.
  • The revenue is real. ~$47B run-rate from enterprise demand separates Claude from circular-financing concerns and makes it the safest platform to build on.
  • Build on the standards, not the model. MCP, Agent Skills, and the zero-rev-share marketplace give you distribution and portability in one move.
  • Keep your moat in data and UX — and hedge the bubble. Ride the winner, price with headroom, and stay portable so a correction can’t take you with it.

Sources

Don't Miss the Next Big Shift

Every week, we break down the trends that matter for indie hackers and SaaS founders — lab announcements, market shifts, and the playbooks the founders ahead of the curve are already running. Stay informed, stay ahead.

Join 3,000+ founders who stay ahead of the curve