AI-First
Anthropic planned to multiply its revenue by 10 in 2026. The company delivered x80. Dario Amodei himself admitted it at the Claude Code developer conference in San Francisco in May 2026. This trajectory, driven almost entirely by a single product, puts Anthropic ahead of OpenAI in annualized revenue for the first time.
What interests me about this story isn't the financial record. It's what it reveals about what businesses actually want: not chatbots, not "digital transformation" in PowerPoint decks, but tools that plug into existing workflows and deliver measurable results.
- 📈 x80 growth: Anthropic went from 1 to 30 billion in annualized revenue in 15 months.
- ⚡ Claude Code, the sole engine: a single product accounts for the bulk of the growth.
- 🏗️ Massive compute crunch: SpaceX deal, Amazon ($25B) and Google ($40B) to keep up.
- 🎯 Clear SMB signal: demand is exploding for AI embedded in processes, not for chatbots.
Here's what Anthropic's numbers really tell us, why Claude Code concentrates all the momentum, and what it concretely changes if you're running an SMB in 2026.
From 1 to 30 billion in 15 months: the trajectory that breaks every model
No tech company had ever reached 1 billion dollars in annualized revenue within four years. Anthropic did it since its founding in 2021, according to BDM. For context: Snowflake and Zoom took a decade, Google five years, OpenAI eight years.
But what came next shattered every benchmark.
How did Anthropic overtake OpenAI in a single quarter?
According to data compiled by tech-insider.org, Anthropic's annualized revenue went from 9 billion at the end of 2025 to 14 billion by mid-February 2026, then 19 billion by end of February, reaching 30 billion in April 2026. Meanwhile, OpenAI reported 25 billion annualized over the same period.
This overtaking matters because it happened in weeks, not years. Anthropic now counts over 300,000 business customers, and more than 500 clients each spend over one million dollars per year with Claude.
| Metric | End of 2025 | April 2026 | Trend |
|---|---|---|---|
| Annualized revenue ($B) | 9 | 30 | ↑ x3.3 in 4 months |
| Enterprise clients | ~200,000 | 300,000+ | ↑ +50% |
| Clients > $1M/year | ~70 | 500+ | ↑ x7 |
| Valuation ($B) | ~180 | 380 | ↑ x2.1 |
| OpenAI annualized revenue ($B) | ~20 | 25 | ↑ +25% |
SOURCE: tech-insider.org, BDM, CNBC · Updated 05/2026
The $30 billion Series G raise at a $380 billion valuation, led by GIC (Singapore's sovereign wealth fund) and Coatue Management, confirms that institutional investors now treat Anthropic as a GAFAM peer, not as a startup.
Claude Code: the product that concentrates all the momentum
I've been helping SMBs adopt Claude Code over the past few months, and what I see on the ground confirms exactly what Anthropic's numbers show at the macro level. Demand isn't for a conversational chatbot. It's for a tool that plugs into real workflows and produces code, automations, and working systems.
Why does a coding tool generate this much revenue?
Dario Amodei said it plainly at the San Francisco conference: the bulk of this x80 growth comes from a single product, Claude Code. The explanation is straightforward. A chatbot consumes a few tokens per conversation. A developer (or a non-developer building with Claude Code) consumes millions of tokens per session.
CNBC reports that Anthropic saw "exponential" growth in Amodei's own words, adding that the company had "planned for anything from it only grows a little to it grows 10X, and yet we saw 80X." This gap between forecasts and reality caused outages, rate-limit reductions during peak hours, and a frantic scramble for compute.
This isn't trivial for SMB leaders. When I see 15-person companies using Claude Code to automate their SEO or build internal dashboards in a few hours, I understand why token consumption is skyrocketing. AI embedded in business processes consumes 10 to 100 times more than a chatbot, and that is precisely where value is created.
OpenAI got the memo too. The company launched Codex, acquired the OpenClaw founder, and invested 1 billion dollars in the same segment. According to The AI Report, the "race for agentic solutions" is now the primary battleground between the two leaders.
The compute crisis: when growth outpaces infrastructure
With growth 80 times above projections, Anthropic literally ran out of computing power. Regular users felt it: recurring outages, reduced token limits during peak hours, degraded response times. The channel Maestros da IA compared the situation to São Paulo's "rodízio," where traffic is restricted by license plate number during rush hour.
How did SpaceX become Anthropic's GPU supplier?
The deal between Anthropic and SpaceX AI (formerly xAI) is arguably the most unlikely alliance of 2026. Elon Musk publicly called Anthropic "misanthropic" (humanity-hating). Anthropic banned employees from using xAI products for fear that data would be used to train Grok. And yet, SpaceX made Colossus 1, a 220,000 NVIDIA GPU supercomputer located in Memphis, Tennessee, available to Anthropic.
The explanation comes down to three words: complementary resources. Anthropic has the best researchers and data but lacks compute. xAI has the compute (Colossus 1 was running at a third of capacity after the migration to Colossus 2) but lacks researchers and quality data. According to Musk himself after spending a week with Anthropic's senior team: "Nothing triggered my bullshit detector."
This deal sits alongside a series of massive agreements signed within weeks:
Amazon invested $25 billion in Anthropic. Google committed $40 billion ($10B in cash plus $30B tied to performance milestones) with 5 gigawatts of Cloud capacity over 5 years. According to Statista estimates, Anthropic is expected to spend over $100 billion in compute over the lifetime of these agreements.
What signal does this "scramble for compute" send to the market?
The signal is clear: demand for AI integrated into operations massively outstrips infrastructure supply. This is no longer a research-lab problem. It's an industrial bottleneck that directly affects the businesses using these tools.
When Anthropic cuts token limits during peak hours, it's the developer at a small company in Lyon who watches their AI agent crash in production. I've experienced this firsthand with my own systems running on Claude. Automations running 24/7 eventually hit these ceilings, and you need to plan for fallbacks.
What Anthropic's growth changes for SMBs in 2026
The temptation, looking at these sky-high numbers, is to think this only concerns large enterprises. It's exactly the opposite.
Should you wait for prices to drop before adopting Claude Code?
No. And here's why. Anthropic's x80 growth is built on companies using Claude Code as a production tool, not as a curiosity. Anthropic's 300,000 business customers aren't all Fortune 500 companies. A significant share are SMBs, freelancers, and tech teams of 5 to 50 people who understood that the real value isn't in the model itself, but in integrating it with business processes.
What I see with my own clients confirms this dynamic. The companies extracting the most value from Claude aren't the ones with the biggest budgets. They're the ones that identified a specific, measurable use case and connected it to their existing tools: CRM, database, back-office, email. If you're looking for concrete integration examples, the GoLive Software blog regularly documents B2B use cases.
How does the OpenAI/Anthropic competition benefit businesses?
The rivalry between the two leaders creates direct leverage for users. OpenAI launched Codex, cut API prices, and offered free fine-tuning to attract developers. Anthropic is investing heavily in infrastructure to stabilize its service.
The concrete result for an SMB leader: per-token prices are falling, capabilities are expanding, and reliability is improving. In Q1 2026, the cost per million output tokens on Claude Sonnet dropped below $3, compared to over $15 a year ago. This cost compression makes AI automations profitable even at modest volumes.
"AI only creates value when it's connected to the company's real tools. Anthropic's $30 billion doesn't come from chatbot conversations. It comes from businesses that plugged Claude into their operations."
Vincent, May 2026
The verdict: what these numbers tell us about the future of AI in business
Anthropic's x80 growth is no accident. It validates a thesis I've been defending since the launch of AI First: businesses don't want AI as a gimmick. They want tools that integrate into their workflows, automate costly tasks, and deliver quantifiable results.
Claude Code captured this demand because it does exactly that. No impressive but useless demos. No 18-month "AI transformation" projects. A tool that installs, codes, automates, and saves time from week one.
For SMBs, the message is simple. Stop burning money on tasks that AI can do better, faster, and cheaper. Anthropic's numbers prove that hundreds of thousands of companies have already figured this out. The question is no longer "should we jump in?" but "which process do we start with?"
My advice: identify the most repetitive and costly task in your week. Test Claude Code on it for 5 days. Measure the time saved. That's how 300,000 companies got started, and that's why Anthropic is reporting $30 billion in revenue.
Frequently asked questions
Is Anthropic really bigger than OpenAI in 2026?
In terms of annualized revenue, yes. Anthropic reached $30 billion in April 2026 versus $25 billion for OpenAI over the same period. This lead is recent (dating from Q1 2026) and the competition remains tight. OpenAI retains a larger consumer user base with ChatGPT, but Anthropic dominates the B2B and developer segment.
Why does Claude Code generate so much revenue for Anthropic?
Claude Code consumes far more tokens per session than a standard chatbot. A developer building a complete system can use millions of tokens in a single day, compared to a few thousand for a typical conversation. This usage intensity, multiplied across hundreds of thousands of professional users, accounts for the bulk of the x80 growth.
Is the Anthropic-SpaceX deal sustainable?
The deal involves the use of Colossus 1, the 220,000-GPU supercomputer in Memphis. SpaceX is migrating to Colossus 2 and had excess capacity. Its longevity will depend on how the relationship between the two companies evolves, but their interests are aligned: Anthropic needs compute, SpaceX monetizes an underutilized asset. Both share a common competitor in OpenAI.
Can an SMB benefit from Claude Code today despite the capacity issues?
Yes. Capacity issues primarily affect peak American hours. Anthropic is investing heavily to resolve them ($65 billion in infrastructure agreements signed within weeks). For a European SMB, morning time slots in the European timezone generally offer stable capacity. The key is to start with a specific use case rather than trying to automate everything at once.
Will Anthropic go public?
According to reports from Bloomberg Television, Anthropic was considering an IPO as early as October 2026 on the Nasdaq. The $30 billion Series G raise may have pushed back that timeline, as private markets already provide the necessary liquidity. The public benefit corporation structure also complicates a standard IPO setup.
Vidéos YouTube
- The situation at Anthropic is INSANE. · Maestros da IA
- Anthropic in Talks to Raise $30 Billion at a $900 Billion Valuation · Bloomberg Television
- Anthropic's strong growth creates challenges for executives · CNBC Television
- Anthropic Beats OpenAI + Google's $40B Bet | AI News in 5 · The AI Why with Liam Lawson