The numbers coming out of Anthropic right now are almost hard to process. The Claude maker is looking at a second-quarter revenue figure that blows past $11.5 billion—a 14-fold jump from the $787 million it booked in the same period last year, according to internal documents reviewed by Bloomberg News. Even more striking: the company posted its first positive adjusted operating income, a milestone it desperately wanted to hit before stepping onto the public market stage. Here's the setup: Anthropic confidentially filed its S-1 with the SEC on June 1. The company is working with Morgan Stanley, Goldman Sachs, and JPMorgan Chase. Six of its backers are telling the Financial Times to expect an October listing at a valuation of $2 trillion or more—which would top SpaceX's $1.77 trillion debut from June and become the largest IPO in history. But as with everything in the frontier AI world right now, there's a gap between what the market is pricing and what the financials actually show.
The Revenue Trajectory Is Not Normal
Let's put this in plain terms. Anthropic ended 2025 with an annualized revenue run rate of roughly $9 billion, according to the company's own disclosures. By May, that figure had crossed $47 billion. Investors are now modeling a year-end run rate between $100 billion and $120 billion. That's a tenfold increase over the course of a single year, and the Q2 numbers bear it out. The $11.5 billion quarterly figure actually beat earlier internal projections of $10.9 billion. It also implies a run rate of about $46 billion—consistent with the $47 billion mark the company cited in May. The growth isn't just coming from one place. Anthropic's enterprise-focused strategy has become the dominant driver. According to Ramp's July AI Index, Anthropic now leads in paid business adoption at 43.5% of eligible companies in its dataset, edging past OpenAI's 39.7%. Around 70–80% of Anthropic's revenue comes from APIs and enterprise contracts billed by tokens, not from individual subscriptions. And Menlo Ventures data suggests Anthropic has overtaken OpenAI in enterprise LLM spending share, with 32% versus OpenAI's 25%—a dramatic reversal from just two years ago, when OpenAI held a 50% share. Then there's Claude Code. Anthropic's developer product has become a revenue monster. Its annual recurring revenue went from $500 million in September 2025 to $2.5 billion by February 2026—a 5x jump in under six months. That now contributes close to 20% of total revenue, with over half of Claude Code's income coming from enterprise subscriptions. According to the company's own metrics, Claude Code now authors roughly 4% of all public GitHub commits globally. One analysis says 71% of AI agent users lean on it as their primary tool. Developers are spending an average of 20 hours a week inside the tool. Enterprise wins also show up in the customer list. More than 1,000 companies now spend upwards of $1 million annually with Anthropic. Eight of the Fortune 10 are Claude customers. Salesforce, Netflix, Spotify, KPMG, L'Oréal, and Novo Nordisk have all been identified as clients. Cognizant has deployed Claude to its 350,000 employees. In financial services, RBC Wealth Management is using Claude inside Salesforce's Agentforce, and Anthropic has a partnership with the UK's FCA to provide Claude models to 21 fintech companies for fraud detection and compliance automation. That kind of customer concentration brings its own risks—Amazon, Google, and Salesforce together reportedly account for around 60% of revenue—but for now, the market is treating it as proof of product-market fit rather than a red flag.
The Road to the IPO
Anthropic's path to the public market has been aggressive. In February, the company raised $30 billion at a $380 billion post-money valuation. By May, it closed a $65 billion round at $965 billion. A $2 trillion IPO would double that valuation in a matter of months. The S-1 filing in June started the clock. CFO Krishna Rao has begun preliminary investor education meetings—described by sources as informal, not a formal roadshow. Executives have reportedly declined to fix a valuation target even in private conversations, which suggests either genuine uncertainty or deliberate poker-playing. On the cost side, Anthropic has backed up its growth ambitions with equally massive commitments. More than $100 billion to AWS over 10 years. $30 billion to Microsoft Azure. Around $200 billion to Google Cloud over five years for TPU compute and cloud services, per a May agreement. Nvidia has committed up to $10 billion in investment. Anthropic also signed a $10 billion six-year deal with Volta Infra for data center capacity in Norway. The revenue quality is improving too. Gross margins on inference have swung from negative 94% a year ago to a range of 70–85%. Compute cost per dollar of revenue has dropped from 71 cents in Q1 to 56 cents in Q2. The company is reportedly approaching a 44% gross margin, trending toward a better unit economics story than ChatGPT maker can currently show, according to PitchBook's AI business quality framework. Meanwhile, Anthropic reportedly expects to achieve adjusted operating profit of $559 million in Q2—a margin of roughly 5.1%, excluding stock-based compensation. That's progress, except for what it implies at scale. A 5.1% operating margin on a $100–120 billion revenue run rate yields roughly $5–6 billion in annual operating profit. At $2 trillion, that's about 300–400x operating profit — a number that's not obviously sustainable without significant margin expansion.
The Valuation Debate: Two Very Different Calculations
The bull case is straightforward: growth this steep deserves aggressive multiples. One investor told the Financial Times that at 800% annual growth, even a conservative 30x revenue multiple gets you to $3 trillion. CNBC's Jim Cramer has defended the valuation, arguing that revenue growth justifies the multiple. Gavin Baker of Atreides Management went further, saying on the All-In Podcast that he thinks Anthropic is "worth $3 trillion today." For reference, Palantir and Nebius have traded at roughly 55x revenue this year. A 30x multiple on $100–120 billion in forward revenue would give you a $3–3.6 trillion outcome. Viewed against those numbers, $2 trillion might actually look reasonably priced. The bear case is equally clear. $2 trillion is a valuation on the scale of Amazon, which just reported $200.6 billion in Q2 revenue and $62.6 billion in net income. Fortune's analysis makes this point: at Nasdaq 100 valuation multiples, Anthropic would need to generate roughly $59–79 billion in annual profits to keep pace. Right now, it hasn't even reported positive net income. FA Magazine cut through the media hype with a blunt reminder: "The awkward part of all this is that Anthropic isn't making money yet." Operating profit is different from net income. For a company with Anthropic's capital intensity, that gap can be huge. Reuters Breakingviews noted the same problem from a different angle: even at a generous 30% operating margin, a $2 trillion valuation would work out to about 56x operating profit on a $120 billion revenue run rate. The tension is whether this is a classic case of paying up for quality growth, or something closer to what Hacker News commenters remember from the dot-com era. One user put it this way: "People started inventing new metrics to 'value' dot-com companies that lost money hand over fist. My favorite was 'revenue multiples.' So instead of a P/E ratio, it was just a multiple of revenue no matter how much money you lost." Another HN user questioned the user base math more directly: "Valuations in this sector have nothing to do with facts anymore, because otherwise no one would give Anthropic a valuation in the same order of magnitude as OpenAI for a tiny fraction of the user base while all objective benchmarks put the top models so close together." On Reddit, developers are looking at this from a different angle entirely. One analysis noted that an IPO filing "would imply something about Anthropic's maturity, disclosure discipline, and the pressure it may soon face to explain its business in public terms." The deeper concern for those building on Claude Code is simpler: does this change release velocity, pricing discipline, and the ability to keep investing in model quality? Another developer pointed out the real-world complications: "At my company the grassroots advocacy from devs has certainly been for Claude Code. Unfortunately even though we have a degree or two of separation from most federal contracts the punitive DoD blacklisting had enough of a chilling effect on our legal team to make them drag their feet on approving any contract involving Anthropic."
The Risks Nobody Wants to Talk About
Anthropic's growth story isn't without turbulence. The Pentagon designated the company a supply-chain risk earlier this year over its refusal to remove restrictions on autonomous weapons and surveillance of Americans. The Commerce Department imposed export controls on Fable 5—Anthropic's newest frontier model—in June, forcing a global suspension until restrictions were lifted. Two investors told the Financial Times this contributed to slower revenue growth in June. There was also the April incident where Anthropic accidentally deleted about 8,100 GitHub repositories while trying to contain a Claude Code source code leak. One account described it as "another black eye for the company as it reportedly plans an IPO," drawing scrutiny over internal controls. Competition is intensifying, too. OpenAI reported annualized revenue north of $40 billion and is preparing its own IPO. DeepSeek, the Chinese lab, is gearing up to go public as early as this year, a looming competitive threat. Anthropic's flagship model costs roughly 2.5x what OpenAI charges for comparable output, and the pricing pressure from Chinese providers like DeepSeek (whose V4 Pro output pricing is reportedly a fraction of Claude's) led Anthropic to cut mid-tier model prices in late July, including pricing Opus 5 at half the rate of Fable 5. Steve Eisman, the "Big Short" investor, has been warning about AI concentration risk, noting that the entire boom relies heavily on just two companies: OpenAI and Anthropic. If OpenAI's next model slips, or if Anthropic hits a security crisis or regulatory blow, both suffer—they're intertwined rather than independent bets.
[SPONSORED]
NEXT-GEN NPU CHIPSETS
Empower your local devices with desktop-class inference capabilities.
The Broader Wave
Anthropic's IPO is one part of a historic wave. SpaceX already went public in June at $1.77 trillion, raising $75 billion. OpenAI filed confidentially around May 22 and could debut as early as September. AI listings this year have raised $256.4 billion, the most since 2021, according to data cited by Reuters—Wedbush analyst Dan Ives called it "a full-on race to Wall Street." Timing matters. Anthropic is attempting to go public while the revenue curve is still steep, but before the full weight of competition, margin pressure, and government scrutiny lands. The index exclusion decision adds another wrinkle: S&P Dow Jones Indices decided in June not to fast-track AI companies into the S&P 500, meaning Anthropic, SpaceX, and OpenAI won't join the index this year, potentially limiting passive fund flows early on. The company is reportedly looking at unconventional liquidity arrangements, too—including expanding 10b5-1 trading plans to all employees to mitigate insider trading risk, and carefully structuring how much stock existing shareholders can sell on day one.
What Happens Next
The IPO valuation won't be set until the company commits to a range in its final S-1 amendment. Until then, the estimates from investors and analysts are simply that—estimates. One investor told the Financial Times that Anthropic's executives "have yet to fix the valuation target for the IPO, even in private conversations." What is known: CFO Krishna Rao is meeting with prospective public-market investors. The company is expected to advance through SEC review through Q3 and into a Q4 roadshow. The growth story is real, and the numbers are real. What's uncertain is whether a company that just barely turned adjusted operating profit positive can justify pricing that puts it in the same league as Amazon. One could argue the smart money is being careful precisely because the numbers are so good. Everything looks perfect at the peak of a growth sprint—it's only after the IPO that the slower, harder work begins. Anthropic's Q2 results are monumental. The question nobody can answer yet is whether they're monumental enough to hold up a $2 trillion valuation when the quarterly reporting cycle starts, and every miss will be measured in headlines.