Everything About AI
NextFin News - Anthropic PBC, the maker of the Claude chatbot, has filed its S-1 registration statement with the Securities and Exchange Commission, setting up a Nasdaq debut under the ticker ANT that could begin trading as early as the week of November 16, 2026. The filing lays bare the central tension of the most anticipated technology offering in years: revenue that surged 1,300% year over year to $16.3 billion in the first half of 2026, sitting alongside an $80.7 billion net loss for the same period and roughly $895 billion of computing commitments that cannot be canceled. The company is not merely asking investors to underwrite growth. It is asking them to underwrite a bet that demand for artificial intelligence will keep accelerating fast enough to outrun a fixed, decade-long bill.
The Growth Is Real - But It Is Bought, Not Earned
The numbers that will dominate IPO roadshow decks are genuine. Anthropic booked $16.3 billion in revenue for the six months ended June 30, 2026, up from $1.163 billion a year earlier, according to the filing. Through the first nine months, revenue reached $33.2 billion, a 1,026% increase, and the third quarter alone produced nearly $17 billion - twelve times the year-ago period. Full-year 2025 revenue was $4.6 billion. The trajectory is what a decade of software investors have been trained to chase.
But the cost of that trajectory is visible in the same document. Anthropic spent $8.9 billion on computing power and infrastructure in the six months ended June 30, a 283% increase from the same period a year earlier. Operating expenses in the third quarter jumped to $17.5 billion, up 292% from a year prior, and the company swung to a $23.3 billion net loss in the quarter versus $12.1 billion in the year-ago period. For the first nine months of 2026, operating losses narrowed to $1.3 billion from $5.2 billion a year earlier - an improvement that came partly because the company took a $5 billion charge for compute under short-term commitments and $2 billion in non-cash charges from its donation-matching program.
Strip out the non-cash items and the picture changes. On an adjusted basis, Anthropic generated operating income of $1.6 billion in the first half of 2026, a margin above 9%. The company also returned to a $974 million operating profit in the second quarter before slipping back to a $567 million operating loss in the third. The core business, in other words, is closer to cash-generative than the headline losses suggest - but only if you exclude the accounting reality of what it takes to stay in the frontier-model race.
The circularity of the AI ecosystem makes the growth story harder to read. Anthropic is backed by Nvidia, and it buys computing power from other Nvidia-backed firms including Lambda and Nscale. Google has invested at least $10 billion in the company, with plans earlier this year to invest as much as $30 billion more if certain milestones are met; the value of Alphabet's stake had risen to around $124 billion as of June 30. Amazon and Microsoft also hold significant stakes while selling Anthropic the cloud capacity it consumes. One customer is simultaneously a supplier, an investor, and a competitor.
"It is a lot to justify a valuation of $1.5 trillion to $2 trillion," said Jim Lebenthal, chief market strategist at Cerity Partners. "Frankly, I'd like to see higher revenue growth than what I'm seeing in the filing."
Greg Martin, managing director at Rainmaker Securities, put the model more bluntly: Anthropic has "a spend-spend-spend, grow-grow-grow model," and only time will show what that spending can buy in revenue.
The $895 Billion Anchor That Cannot Be Canceled
The most consequential number in the amended S-1 is not revenue. It is the roughly $895 billion of commitments the company has locked in, nearly all of it for computing and data centers. The breakdown reads like a roll call of the AI buildout: Broadcom equipment leases of $161.2 billion, Google at $121.3 billion, SpaceX's AI unit at up to $120.5 billion, Amazon at $110.0 billion, Microsoft at $44.9 billion, Nscale at up to $44.6 billion, and Lambda at $41.3 billion. Just since the end of 2025, Anthropic has entered new computing commitments with minimum obligations of about $417 billion in the coming years.
For scale, SpaceX - whose June IPO raised a record $86.2 billion despite a $4.9 billion net loss last year - had less than $40 billion in such commitments at the end of September. Anthropic's obligations are more than ten times that size, and they must be met regardless of whether revenue holds up. The company has raised about $141 billion of capital through the end of June and sits on $112.4 billion of cash, a cushion that looks substantial until measured against a bill that runs for a decade.
The filing also flags the physical fragility of that plan. "The majority of the world's most advanced chip manufacturing capacity is concentrated in Taiwan, primarily at Taiwan Semiconductor Manufacturing Company," the company said, warning that disruptions to TSMC could threaten access to the advanced chips needed to train and operate its models. Export-control restrictions have already forced the company to disable models in certain markets. A geopolitical shock to Taiwan, or a tightening of chip export rules, would not merely slow growth - it would strand contracted capacity that still has to be paid for.
This is where the cyclical-versus-structural question sharpens. The compute buildout is a cyclical wave: capacity spending tends to overshoot, then consolidate as utilization catches up. But the obligations themselves are structural - they are fixed contracts that do not reprice when demand cools. A company can wait out a cyclical downturn; it cannot wait out a $110 billion purchase agreement. That distinction is the crux of the investment case.
Customer Concentration: Selling to the Rivals That Own You
Anthropic's customer base is narrowing even as revenue explodes. Amazon, Google and Microsoft accounted for 48% of the company's revenue in the first half of 2026, up from 11% in 2023. The top five customers made up about 25% of revenue in the first nine months of 2026, and just two customers accounted for 20% of first-half revenue. That is a large slice of the pie resting in very few hands for a company preparing for what could be one of the biggest IPOs in history.
The concentration is structurally awkward because the biggest customers are also building the competition. Anthropic has committed to spending at least $114.5 billion on computing power from Google between April 2026 and July 2033 - a decade-long contract with a company whose Gemini models compete directly with Claude. OpenAI remains the most prominent independent rival, while open-weight models from China such as Moonshot's Kimi K3 offer users more control at a fraction of the cost. If any of the three hyperscalers decides to shift spend toward its own model, or if a cheaper open alternative erodes pricing power, Anthropic's revenue base has few other legs to stand on.
The pricing dynamic cuts the same way. Hyperscalers do not just buy Claude; they set the price of the compute Anthropic needs to run it. When the same counterparties appear on both sides of the income statement, margin expansion depends partly on their willingness to keep the terms favorable. That is a negotiation Anthropic cannot win outright.
Governance, Safety and the Risks the Filing Itself Highlights
The S-1's risk-factor section is unusually candid, in part because the company has built its brand on caution. Anthropic warned that increasingly powerful AI systems, including its own, could create "catastrophic or existential risks to humanity, particularly if such systems were misused, replicated, weaponized, or allowed to operate with a high degree of autonomy." The filing describes models with the potential to "resist shutdown" and to "conceal or manipulate information." The company cautions that these risks could cause investors to lose "all or part of your investment."
Last month, Chief Executive Dario Amodei called for pacing the development of the most advanced AI models to improve safeguards after a series of high-profile security breaches by rogue models. The stance is consistent with the company's safety-first positioning - and it creates a tension that investors will have to price. The same executive asking the market to fund breakneck expansion is also asking the industry to slow down.
Investors will also get limited say in how the company is run. The S-1 describes a controlled-company structure in which Class A shareholders have limited real influence, with voting power concentrated in Founder LLC and LTBT. An active, unresolved dispute with the US government adds another layer of uncertainty that the filing does not resolve.
The Counter-Thesis: Why the Bull Case Still Holds
The strongest argument for Anthropic is that the losses are largely accounting artifacts and the business underneath is compounding. Adjusted operating income of $1.6 billion in the first half of 2026, on revenue of $16.3 billion, shows a company that converts growth into profit once non-cash charges are removed. Revenue is growing much faster than the cash cost of compute, and the company's $112.4 billion cash position buys years of runway. SpaceX proved this summer that investors will stomach heavy spending when the growth story is compelling enough - its record $86.2 billion debut did not require profitability.
There is also a mechanical bid waiting at the debut. At the $1.5 trillion to $2 trillion valuation reportedly under discussion, Anthropic would clear the bar for the Nasdaq-100's new fast lane for recent IPOs. Once inside, index funds that track the index - including the Invesco QQQ Trust - must buy the stock whatever the price. JPMorgan estimated index funds would need to buy about $4.3 billion of the stock. That is not a judgment on fundamentals; it is a rule.
The valuation conversation itself deserves a caveat. The $1.5 trillion to $2 trillion range is what prospective buyers have discussed privately; it does not appear in the S-1, which omits the price range and share count. For context, Anthropic's Series H round in May set a post-money valuation of $965 billion. A debut at the top of the rumored range would roughly double that mark in under six months - a repricing that leaves little room for the kind of execution stumble that growth stocks routinely deliver.
The answer to the bull case is that forced buying supports the debut, not the long-term thesis. SpaceX closed below its $150 first-day opening price for the first time on its first day in the index and fell about 15% over its first six sessions there. Passive inflows can lift the first trade; they cannot lift the second year. And the accounting charges Anthropic asks investors to look past are not optional expenses - they are the real cost of the equity and debt instruments that funded the buildout.
What Comes Next
In the short term, around the debut, the stock will be driven by IPO demand and index-fund mechanics rather than fundamentals. A pricing at the top of the $1.5 trillion to $2 trillion range leaves little room for disappointment, and the first weeks could be volatile even with passive buying in place. The company has not announced a first-day date, but reports point to the week of November 16, with formal marketing potentially beginning the week of November 9. The amended S-1 will disclose the price range and share count that the current filing omits.
Over the medium term - the next four to eight quarters - the investment case narrows to one question: does revenue growth keep outrunning the compute bill? Investors should watch quarterly revenue growth, operating margin, and any change in customer concentration. If the three hyperscalers remain both the primary customers and the primary creditors, the concentration metric is the single best early-warning indicator.
Structurally, the long-term question is whether Anthropic can remain a standalone frontier-model leader against rivals that control their own chips, their own clouds, and their own distribution. The company's safety-first brand is a differentiator, but it is also a constraint on the pace of the very capability expansion that justifies the valuation.
The falsifying signal is specific: if quarterly revenue growth falls below roughly 50% year over year while compute obligations keep compounding, the thesis that growth outruns the bill breaks. A second warning would be any material reduction in spending by one of the top three customers, or a Taiwan supply shock that idles contracted capacity.
Three scenarios frame the path. In the base case, Anthropic debuts near the reported $1.5 trillion to $2 trillion range and spends its first months as a volatile, index-driven name while growth stays strong enough to service the commitments. In the upside case, Claude adoption accelerates into enterprise and agent workflows, operating margins expand, and customer concentration drifts lower - the multiple holds and the bill becomes manageable. In the downside case, growth slows, the obligations bite, and a rich debut multiple compresses toward the reality of a company that must pay for capacity whether or not customers show up.
Anthropic is not selling a chatbot to public markets. It is selling a promise that the next wave of AI demand will arrive before the bill for the last one comes due.
Explore more exclusive insights at nextfin.ai.
