Anthropic raises 65 billion dollars in Series H and launches Claude Opus 4.8: the day everything changed
🔎 65 billion in one morning, a model at the top, and OpenAI relegated to second place
On May 28, 2026, Anthropic pulled off a double whammy that shook Silicon Valley. On the one hand, a 5 billion dollar Series H funding round bringing the post-money valuation to 65 billion dollars — an absolute record for an AI startup. On the other, the release of Claude Opus 4.8, which takes the top spot on the Artificial Analysis Intelligence Index with a score of 61.4, dethroning OpenAI's GPT-5.5.
This is the first time an AI startup has surpassed OpenAI in valuation. And it is the first time Anthropic has placed a model at the absolute top of a benchmark ranking. All on the same day.
According to TechCrunch, the round was closed in less than three weeks. Le New York Times speaks of a "tipping point" in the AI race. Anthropic is no longer playing the role of cautious challenger. It is taking the lead.
The key points
- Anthropic raises 5 billion dollars in Series H at a post-money valuation of 65 billion dollars (May 28, 2026), led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital.
- Claude Opus 4.8 takes first place on the Artificial Analysis Intelligence Index with 61.4 points, ahead of GPT-5.5 (OpenAI) and Gemini 3.1 Pro (Google).
- Revenue run-rate reaches 7 billion dollars, a spectacular increase compared to the early signs of quarterly profitability observed earlier in the year, recalling Anthropic's profitability trajectory.
- Anthropic surpasses OpenAI as the most highly valued AI startup in the world.
- An IPO is expected for October 2026, with a valuation target that could approach 100 billion dollars.
Recommended tools
| Claude Opus 4.8 | Flagship LLM, advanced reasoning | Quote-based (June 2025, check on anthropic.com) | Enterprises, complex R&D |
|---|---|---|---|
| GPT-5.5 | Generalist LLM #2 | Quote-based (June 2025, check on openai.com) | Development, automation |
| Gemini 3.1 Pro | Google multimodal LLM | Quote-based (June 2025, check on ai.google) | Google Cloud ecosystem |
| Claude Code | Anthropic coding agent | 20-200$/month (June 2025, check on anthropic.com) | Developers |
Series H: $5 billion in three weeks
Five billion dollars. In a single round. Led by four funds that don't often get it wrong: Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital.
The amount is colossal, but it's the speed that is truly staggering. According to Anthropic's official announcement, discussions began in early May. The term sheet was signed on the 20th. The money was in the accounts on the 28th.
This speed reflects a strong conviction among investors: Anthropic has a real, not hypothetical, competitive advantage. The $7 billion run-rate is not a distant projection. It is the figure the company is displaying at the time of the raise.
To put things into perspective, $65 billion is more than Schneider Electric's market capitalization in early 2025. It's the GDP of a country like Croatia. And it's driven by a company with fewer than 3,000 employees.
This round is part of a logic of continuous scaling, following the $30 billion round that already shook the sector. Except here, we are no longer talking about $30 billion, but $65 billion. Double in just a few months.
Why a Series H and not a direct IPO?
The answer is pragmatic. Anthropic wants to go public with an irrefutable growth story. The Series H serves as a springboard: it locks in the valuation at $65 billion, gives the company massive cash reserves (more than $12 billion in total), and buys time to prepare the books for an October 2026 IPO.
A rushed IPO would have meant a valuation negotiated under market pressure. The Series H, on the other hand, is negotiated behind closed doors with long-standing partners. Anthropic controls the narrative.
Claude Opus 4.8: the game-changing model
On the very day of the funding round, Anthropic releases Claude Opus 4.8. The timing is no coincidence. It shows investors that the company isn't just raising money — it's also delivering the best model on the market.
According to FelloAI, Claude Opus 4.8 scores 61.4 on the Artificial Analysis Intelligence Index. GPT-5.5, which had been dominating for months, drops to second place. Gemini 3.1 Pro (Google) follows at 92 points on the overall LLM ranking, but the Artificial Analysis composite index weights capabilities differently.
The leap isn't just cosmetic. Claude Opus 4.8 significantly improves reasoning on agentic tasks, long context management, and the reliability of structured outputs. These are areas where Claude Opus 4.7 (Adaptive) was already excellent in agentic (94.3 points), but where Opus 4.8 takes it to the next level.
AWS immediately deployed the model on Bedrock, meaning instant access for enterprises using the Amazon ecosystem. It's a strong signal: the Anthropic-AWS partnership remains the cornerstone of enterprise distribution.
What sets Opus 4.8 apart from the competition
The difference isn't measured solely in benchmark points. Claude Opus 4.8 introduces more robust context management over 200K+ token windows, with fewer hallucinations in the middle and end of the context. It's the kind of progress that is invisible in a composite score but changes the lives of developers.
For coding tasks, Opus 4.8 outperforms GPT-5.3 Codex in test generation and refactoring. It naturally ranks among the best LLMs for coding, cementing Anthropic's position in this strategic niche.
The comparison with ChatGPT also becomes more clear-cut. In the Claude vs ChatGPT face-off, Opus 4.8 reverses the dynamic: it is now Claude that holds a clear advantage on complex tasks, while ChatGPT/GPT-5.5 retains a slight edge in mainstream ease of use.
From 0 to a $7 billion run-rate: the fastest growth in tech history
The number that hurts OpenAI the most is the run-rate. Seven billion dollars in annualized revenue at the time of the Series H.
Just five months ago, Anthropic achieved its first quarterly profit. Anthropic's first profitability was seen as a positive but modest signal. No one predicted such a brutal acceleration.
Two factors explain this explosion. First, enterprise contracts. Anthropic has signed multi-month deals with Fortune 500 companies that sometimes exceed $100 million each. The enterprise joint ventures launched by Anthropic and OpenAI show that the enterprise market is the real battlefield, and Anthropic is rapidly gaining market share there.
Second, the agentic pivot. Companies are no longer paying just for chat. They are paying for agents that execute complete workflows. And this is where Claude excels. The recent launch of Claude Code Agent View illustrates this strategy: Anthropic is no longer selling a model, it is selling an autonomous work environment.
The comparison with OpenAI's trajectory
OpenAI took four years to reach a $7 billion run-rate. Anthropic did it in two and a half years. Of course, the market was different — OpenAI educated the demand. But Anthropic's adoption speed remains unprecedented.
The parallel is striking. In 2024, OpenAI raises at an $80 billion valuation and everyone screams bubble. In 2026, Anthropic raises at 65 and those same voices go quiet, because the revenue is there. The difference between a bubble and reality is traction.
The $65 billion valuation: bubble or rationality?
This is the question everyone is asking. And the honest answer is: both.
If we apply a classic SaaS multiple of 10x on the run-rate, we get $70 billion. Anthropic at $65 billion is therefore "reasonable" compared to its current revenues. It's even a slight discount, which is rare in AI.
But this multiple assumes that growth continues. If the run-rate stagnates at $7 billion — because the competition catches up, because companies cut AI budgets, because a new model changes the game — then the valuation collapses. A 10x multiple on revenues that are no longer growing is $70 billion. A 5x multiple is $35 billion. Half.
Le New York Times notes that some private investors already value Anthropic at over $90 billion on the secondary market. This suggests that the Series H at $65 billion could be seen as "cheap" compared to IPO expectations.
What this valuation says about the market
It says that investors are betting on a three-way oligopoly: OpenAI, Anthropic, Google. Not five, not ten. Three. And in an oligopoly, the second player is worth almost as much as the first.
It also says that the enterprise market is the real driver of value. Consumer models generate traffic and brand awareness. But enterprise contracts generate billions. Anthropic understood this faster than OpenAI, which long favored B2C.
The competitive landscape after May 28
Before May 28, the hierarchy was clear: GPT-5.5 in the lead, Gemini 3.1 Pro second, Claude Opus 4.7 third. After May 28, everything becomes blurred.
Claude Opus 4.8 takes first place on the composite index. But on specialized rankings, the situation remains nuanced. In pure agentic, GPT-5.5 maintains its lead (98.2 vs an estimated score around 96 for Opus 4.8). In deep reasoning, Gemini 3 Pro Deep Think (95.4) remains formidable.
The real shift is psychological. For two years, the question was "when will Claude surpass GPT?" On May 28, the answer came. Even if the advantage is slim, even if GPT-5.5 could reclaim the top spot in a month, the symbol is huge.
The Moonshot AI threat and the open-weight world
Anthropic does not dominate everything. Moonshot AI raised 2 billion dollars and its model Kimi K2.6 dominates the open-weight segment with an agentic score of 88.1 in self-host. This is an asymmetric threat: Kimi K2.6 does not directly compete with Claude on the premium enterprise market, but it siphons off developers and startups who want self-hosting.
The table below summarizes the new hierarchy:
| Model | Publisher | AI Index Score | Agentic Score | Status |
|---|---|---|---|---|
| Claude Opus 4.8 | Anthropic | 61.4 (#1) | ~96 (estimated) | Proprietary |
| GPT-5.5 | OpenAI | ~60.8 (#2) | 98.2 (#1) | Proprietary |
| Gemini 3.1 Pro | 92 (general) | 87.3 | Proprietary | |
| Kimi K2.6 | Moonshot AI | 84 (general) | 88.1 (self-host) | Open-weight |
| Claude Opus 4.7 | Anthropic | 90 (general) | 94.3 | Proprietary |
The consequences for OpenAI
OpenAI finds itself in an uncomfortable position for the first time since the launch of ChatGPT. Second in valuation. Second in benchmark scores. And with an IPO that still hasn't been announced.
OpenAI's response will be crucial. They can accelerate GPT-6, they can lower prices, they can multiply partnerships. But they can no longer rely on the "we are the undisputed leaders" narrative. That narrative died on May 28, 2026.
The danger for OpenAI isn't Anthropic in itself. It's that the market is starting to see AI as an oligopoly where positions are fluid. If Anthropic can surpass OpenAI in six months, why not Google in another six months? Investor confidence is cracking.
To follow these movements, the AI trends ranking has become an essential reading tool. Positions change every month.
The infrastructure behind the rise
Claude Opus 4.8 does not exist in a vacuum. Its deployment relies on massively strengthened cloud infrastructure. The partnership with AWS, announced as immediately deployed on Bedrock, gives Anthropic a distribution advantage that neither OpenAI nor Google can easily match.
Anthropic uses AWS trainium and inferentia clusters for training and inference, reducing its dependence on NVIDIA GPUs. This is a major strategic point: less hardware dependency means better margins and more predictable scalability.
For companies that want to deploy their own solutions around these models, hosting infrastructure remains a key factor. Solutions like Hostinger can serve as a foundation for small deployments, but companies using Opus 4.8 at scale go through AWS Bedrock or Anthropic's direct API.
And now? The road to the IPO
Everything that happened on May 28 is designed to prepare the October 2026 IPO. The raise locks in the floor valuation. The model proves technical superiority. Revenues prove commercial traction.
The path is set. Anthropic wants to go public with a valuation between $90 and $120 billion. If the secondary market is already at 90+, an IPO at 100+ is realistic. This would make it one of the largest tech IPOs in history, behind Alibaba (2014) and perhaps Meta (2012).
But the IPO is also a risk. Once public, Anthropic will be judged quarter by quarter. The benchmarks will change every month. A bad score, a lost contract, a model delay — and the valuation will take a hit. Private life was more comfortable.
What the IPO would change for users
Not much on the product front. But potentially a lot regarding transparency. Public companies publish detailed figures. We will know exactly how much Anthropic makes per product (API, Claude Pro, enterprise contracts), how much it spends on compute, and what its real margin is.
❌ Common mistakes
Mistake 1: Confusing valuation and cash
Anthropic is valued at 65 billion. That doesn't mean it has 65 billion in the bank. The Series H brings in 5 billion. Valuation is a market price, not a liquid asset. Confusing the two leads to flawed analyses about Anthropic's ability to "buy" competitors or survive without revenue.
Mistake 2: Believing that #1 on a benchmark is definitive
Claude Opus 4.8 is #1 today on the Artificial Analysis Intelligence Index. GPT-5.5 was #1 yesterday. GPT-6 or Gemini 4 could be #1 tomorrow. The benchmark race is a perpetual sprint. Betting on a permanent leader means ignoring the history of this sector.
Mistake 3: Ignoring AWS's role in the valuation
Anthropic isn't worth 65 billion solely because of Claude. It is worth 65 billion because AWS is its main distributor, Google Cloud is a secondary partner, and the company isn't dependent on a single channel. Distribution infrastructure matters just as much as the model.
Mistake 4: Comparing Anthropic's valuation to OpenAI's 2024 valuation
Market conditions have changed. Interest rates have dropped. AI revenue has exploded. Comparing Anthropic's Series H in 2026 to OpenAI's rounds in 2024 without adjusting for context is comparing apples to oranges.
❓ Frequently Asked Questions
Who led Anthropic's Series H?
The round was co-led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital. These are four top-tier tech funds with strong track records in major tech IPOs.
Is Claude Opus 4.8 really better than GPT-5.5?
On the Artificial Analysis Intelligence Index, yes (61.4 vs ~60.8). In pure agentic tasks, GPT-5.5 remains slightly superior (98.2). The verdict depends on your use case: complex reasoning and long context for Claude, pure agentic tasks for GPT-5.5.
What is the connection between this funding round and the IPO?
The Series H is a pre-IPO round. It locks in a $65 billion valuation, gives Anthropic a comfortable cash position, and prepares for the planned October 2026 public listing with a target of $90-120 billion.
How to access Claude Opus 4.8?
The model is available via the Anthropic API, on AWS Bedrock, and for Claude Pro/Max subscribers. Exact pricing can be found on anthropic.com. The overview of the best AI tools will be updated to reflect this availability.
Is Anthropic really valued higher than OpenAI?
In terms of private startup valuation, yes. At the time of the Series H, OpenAI had not announced a new round exceeding $65 billion. However, OpenAI remains the best-known AI company among the general public and potentially generates comparable revenue.
✅ Conclusion
On May 28, 2026, Anthropic proved that it was possible to raise $5 billion, release the best model in the world, and surpass OpenAI in valuation — all on the same day. This is no longer a rising startup. It is a giant that has taken power. To follow the next steps toward the IPO, keep an eye on the AI trends feed.