Anthropic vs OpenAI Valuation: How the Colossus Deal Pushed Anthropic Past $1 Trillion
Anthropic now implies $1T+ on secondary markets vs OpenAI's $850B. The compute race just reshuffled the AI power rankings.
Anthropic Is Now Worth More Than OpenAI. Here’s Why That’s Not as Crazy as It Sounds.
Anthropic is trading at an implied valuation of over $1 trillion on secondary markets. OpenAI, the company that invented the modern AI race, is sitting at $850 billion. If you had told anyone that in 2023, they would have laughed at you. Yet here we are in 2026, and the valuation gap between these two companies has not only closed — it has flipped.
The question you should be asking isn’t whether the numbers are real. Secondary market valuations are notoriously squishy. The question is what drove this inversion, whether it reflects something durable, and what it means if you’re building on top of either platform.
The short answer: Anthropic solved its most dangerous problem. OpenAI may have created a new one.
The Compute Gap Was Always the Real Story
For most of 2025 and into early 2026, Anthropic was losing the AI race in the most boring way possible: it couldn’t serve its own demand. The models were good — often better than OpenAI’s equivalents on the benchmarks that matter for production use cases — but users kept hitting rate limits, getting throttled during peak hours, and watching quotas shrink without explanation. As one analyst put it, Anthropic had a fantastic model with weak capacity.
This wasn’t a secret. Dario Amodei had made a deliberate decision years earlier to be conservative on capital expenditure, reasoning that if AI demand didn’t accelerate at a perfect rate, over-investing in GPUs would put the entire company at risk. OpenAI took the opposite position: raise everything, buy everything, leverage up. That bet looked right for a long time.
Then Dario disclosed something that reframes the entire narrative. In Q1 2026, Anthropic saw 80x annualized growth in revenue and usage — in a single quarter. They had planned for a world of 10x annual growth. They got eight times that. The conservative capex strategy, which looked prudent in 2023, had become a structural liability by 2026.
The compute shortage was so severe it was actively damaging the product. Anthropic’s compute shortage and its effect on Claude limits had become a real competitive disadvantage, and OpenAI was exploiting it. Every time a developer hit a Claude rate limit, they had a reason to try Codex.
What the SpaceX Deal Actually Changed
The Colossus 1 announcement is the inflection point. Anthropic didn’t just sign a partnership with SpaceX — it took over the entire Colossus 1 data center in Memphis, Tennessee. All 220,000 Nvidia GPUs. All 300 megawatts of capacity. Not a partial lease. The whole thing.
That’s a meaningful distinction. When Anthropic says “we will use all the compute capacity at their Colossus 1 data center,” they mean XAI is not serving anything else from that facility. Elon’s explanation — that SpaceX AI had already moved its own training to Colossus 2, which houses roughly 550,000 Blackwell GPUs — makes the arrangement coherent from his side. The H100s in Colossus 1 were sitting underutilized. Anthropic needed them immediately.
The rate limit changes that followed were immediate and concrete. Claude Code’s 5-hour rate limit was doubled for Pro, Max, Team, and seat-based enterprise plans. Peak hour usage reduction on Claude Code was eliminated for Pro and Max accounts. API rate limits for Opus models were raised substantially: Tier 1 max input tokens per minute went from 30,000 to 500,000; Tier 2 from 450,000 to 2 million; Tier 3 from 800,000 to 5 million; Tier 4 from 2 million to 10 million.
Those aren’t incremental improvements. That’s a 16x increase at Tier 1. For developers who had been routing around Anthropic’s limits using workarounds, this changes the calculus.
And the Colossus deal is just one piece. Anthropic also has an agreement with Amazon AWS for up to 5 gigawatts of new compute, with nearly 1 gigawatt coming online by end of 2026. There’s a 5 gigawatt deal with Google and Broadcom beginning in 2027. And a $30 billion Azure capacity deal with Microsoft and Nvidia. The company went from compute-starved to having more infrastructure commitments than it can immediately absorb.
Why the Valuation Inversion Happened
Secondary market valuations are driven by perceived future cash flows, competitive positioning, and — frankly — narrative momentum. All three shifted toward Anthropic in Q1 2026.
Plans first. Then code.
Remy writes the spec, manages the build, and ships the app.
On cash flows: Anthropic is at $30 billion in annualized revenue, up from $9 billion four months earlier. That’s not a rounding error. That’s the fastest revenue growth of any company in recorded history, faster than any hypergrowth SaaS company you’ve heard of. The 80x annualized growth figure Dario disclosed isn’t a marketing claim — it’s the explanation for why investors are described as having “nearly insatiable” demand for Anthropic shares on secondary platforms.
On competitive positioning: Anthropic has 42 to 54 percent market share in AI coding, which is now 51 percent of all enterprise generative AI usage. Claude Code alone is doing $2.5 billion in annualized revenue. That single product line is larger than most public SaaS companies. The competitive dynamics between Anthropic, OpenAI, and Google on agent strategy have shifted materially in Anthropic’s favor, particularly in the enterprise coding segment where OpenAI has just 21 percent share.
On narrative: The Pentagon story matters here. When the Trump administration designated Anthropic a supply chain risk for refusing to remove restrictions on autonomous weapons use, it should have been catastrophic. Instead, Claude became the number one app in the App Store within hours. Enterprise legal and compliance teams suddenly had a story they could take to their boards. That kind of brand differentiation is nearly impossible to manufacture and very hard to dislodge.
OpenAI, by contrast, is dealing with investor skepticism, a public trial producing unflattering text messages, and a product narrative that has been on the defensive. The GPT 5.5 Instant release — a refined default model, not a new frontier model — is the kind of incremental update that doesn’t move secondary market sentiment.
The Elon Factor: What It Means and What It Doesn’t
The most surreal part of this story is Elon Musk’s role. As recently as March 2026, he was tweeting “Is there a more hypocritical company than Anthropic?” and calling them “missanthropic.” His tweet praising Anthropic’s team — “Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector” — got 20 million views and represents one of the more dramatic public reversals in recent tech history.
The cynical read is correct: this is the enemy of my enemy. Elon is in active litigation with Sam Altman. Helping Anthropic pull ahead of OpenAI is, from Elon’s perspective, a way to damage his primary adversary while also monetizing excess compute capacity that was costing him money to leave idle. XAI’s Grock models haven’t kept pace with Claude or GPT, and the company is being dissolved as a separate entity and rebranded as SpaceX AI. The compute is real. The ideological conversion is not.
Chimath Palihapitiya called this exact dynamic on the All-In podcast weeks before the deal was announced, predicting that Elon would use compute capacity as leverage in exactly this way. That’s not a coincidence — it’s a structural observation about who holds power when physical infrastructure is the bottleneck.
Hire a contractor. Not another power tool.
Cursor, Bolt, Lovable, v0 are tools. You still run the project.
With Remy, the project runs itself.
What the deal does mean, practically, is that Anthropic now has a compute partner with demonstrated ability to build infrastructure fast. Colossus 1 went from concept to operational in a matter of months. Tom Brown, one of Anthropic’s founders, tweeted: “Grateful to be partnering with SpaceX. We’re going to need to move a lot of atoms.” The XAI post also mentioned that SpaceX and Anthropic have “expressed interest in partnering to develop multiple gigawatts of orbital AI compute capacity.” That’s speculative, but it signals the scope of what both parties are thinking about.
OpenAI’s $850 Billion Problem
OpenAI’s valuation isn’t collapsing — $850 billion is still an extraordinary number. But the gap has closed and inverted, and the reasons why matter for anyone building on these platforms.
OpenAI’s infrastructure advantage, which was supposed to be durable, turned out to be less durable than expected. Anthropic closed the gap in months through a combination of deals that would have seemed implausible a year ago. The $200 billion commitment to Google Cloud, the Amazon AWS deal, the Microsoft Azure capacity, and now Colossus 1 — Anthropic went from compute-constrained to having more committed capacity than it can immediately deploy.
OpenAI’s product narrative has also become more complicated. The trial testimony from Mira Murati about Sam Altman’s conduct during the firing episode, the $30 billion stake valuation for Greg Brockman, the text messages — none of this is fatal, but it’s not the story you want circulating when you’re trying to maintain a premium valuation.
The model gap is real too. Claude Opus 4.7 versus GPT-5.5 on coding tasks shows Anthropic maintaining a meaningful lead in the use cases that drive enterprise revenue. And Claude Mythos, Anthropic’s unreleased frontier model, scored 77.8 percent on SWE-bench Pro — roughly 20 points higher than the next best model. Anthropic has two models ahead of the competition simultaneously. That’s not a benchmark artifact; that’s an architectural lead.
For developers building agents and workflows, the practical implication is that Anthropic’s platform is now more reliable than it was six months ago, and the trajectory is improving. Platforms like MindStudio — an enterprise AI platform with 200+ models, 1,000+ integrations, and a visual builder for orchestrating agents and workflows — give you the flexibility to route between providers as capacity and performance shift, which is exactly the kind of optionality that matters when the infrastructure landscape is moving this fast. If you’re building spec-driven full-stack applications on top of these models, tools like Remy let you write a markdown spec with annotations and compile it into a complete TypeScript app — backend, database, auth, and deployment included — so your development velocity doesn’t have to wait on the infrastructure wars to settle.
What the Valuation Gap Actually Tells You
Secondary market valuations are not precise instruments. They reflect sentiment, momentum, and the collective guess of sophisticated investors about where value will accrue. The $1 trillion implied valuation for Anthropic and $850 billion for OpenAI are both subject to revision.
But the inversion is telling you something real: the market has updated its view on which company has the more durable competitive position. Anthropic’s combination of model quality, enterprise adoption, brand differentiation, and now compute capacity has convinced secondary market buyers that the growth trajectory is more defensible than OpenAI’s.
Remy doesn't write the code. It manages the agents who do.
Remy runs the project. The specialists do the work. You work with the PM, not the implementers.
The 80x annualized revenue growth figure is the anchor. If you’re growing that fast and you’ve just solved your primary constraint — compute — the forward-looking math gets very large very quickly. OpenAI is still growing, still building, still releasing models. But the narrative that it was the inevitable winner of the AI race has been replaced by something more contested.
For builders, the practical question isn’t which company has the higher implied valuation. It’s which platform you can actually build on reliably. Six months ago, the answer was complicated by Anthropic’s rate limits. Today, with Colossus 1 online and more capacity coming, that answer is clearer. If you’re building production systems that depend on Claude’s coding capabilities — and Claude Code’s performance relative to alternatives has been a genuine differentiator — the infrastructure story now supports the model story in a way it didn’t before.
The valuation gap is a lagging indicator. The compute deals, the revenue growth, and the enterprise adoption numbers are the leading ones. Those have been pointing in Anthropic’s direction for longer than the secondary market prices have reflected it.
Whether OpenAI can respond — with its own infrastructure moves, with model releases, with the Codex ecosystem — is the actual question worth watching. The AI race is not over. But the scoreboard looks different than it did at the start of the year, and the reasons why are more structural than they might appear.
