Everything About AI
NextFin News - Anthropic's annualized revenue run rate is on a path to top $100 billion before the end of 2026, according to people briefed on the company's investor updates, a trajectory that would hand the Claude creator one of the most aggressive growth stories ever taken public. The figure matters less for what it says about the past five months than for what it demands of the next three years: at the roughly $2 trillion valuation investors are weighing for an initial public offering, the market would be paying about 17 to 20 times a revenue line that barely existed 19 months ago.
The number sits at the center of a debate that will define the AI cycle's next phase. Anthropic's annualized revenue run rate reached $65 billion at the end of July, up from $47 billion in May and about $9 billion at the end of 2025 — a sevenfold jump in a year. Investors now expect the company to finish 2026 with annualized revenue between $100 billion and $120 billion, and to be projecting roughly $190 billion to $200 billion by 2028. The question for public-market investors is not whether the company can sell more API calls. It is whether the revenue curve is a structural rerating of software economics or a cyclical infrastructure boom that will flatten once the industry's compute buildout matures.
The Numbers Behind the Run Rate
Run rate is a projection, not a trailing result: it takes a recent period of sales and multiplies it to a full year. That distinction is the entire point of the Anthropic story. On a calendar-year basis, the company will not book $100 billion in 2026 — by one reconstruction of its disclosed checkpoints, actual 2026 revenue lands closer to the low tens of billions, because the exponential growth is concentrated in the second half. But the market prices forward, and the forward line is nearly vertical.
The progression, as reconstructed from the company's disclosed checkpoints, runs like this. Anthropic exited 2025 at roughly $9 billion in annualized run rate. By February 2026 it was $14 billion, then $19 billion in March, $30 billion in April — a figure the company's executives cited in fundraising discussions that month — and $47 billion by mid-May, disclosed alongside the closing of a $65 billion Series H round that valued the company at $965 billion. The run rate then reached $65 billion at the end of July. The company also shared a preliminary second-quarter revenue figure of $11.5 billion, a 14-fold increase from a year earlier.
For comparison, rival OpenAI's annualized revenue run rate recently reached $40 billion, roughly doubling from about $20 billion at the end of 2025. Both companies have filed confidential IPO paperwork. Anthropic submitted a confidential draft registration statement to the Securities and Exchange Commission on June 1, saying the move "gives us the option to go public after the SEC completes its review," and that any offering "will depend on market conditions and other factors." Betting markets have priced an October debut as the base case, with roughly an 83 percent probability of going public by the end of that month.
What a $2 Trillion Valuation Actually Prices In
The valuation math is where the story sharpens. A $2 trillion listing against a $100 billion to $120 billion year-end run rate implies a multiple of about 16.7 to 20 times revenue. Against the $65 billion July run rate, the same valuation would be a 30.8 times multiple — the kind of ratio usually reserved for software companies growing at triple-digit rates with best-in-class margins.
That multiple only makes sense if three things hold. First, growth must persist: the $190 billion to $200 billion 2028 projection the company has shared with investors implies the run rate roughly triples again in two years. Second, margins must expand: at $65 billion of run rate, the company is spending heavily on compute capacity, and the capital intensity of training frontier models has been the industry's defining constraint. Third, the competitive position must hold: Anthropic has pulled ahead of OpenAI on reported run rate, but the lead is measured in months, not years, and both labs are racing to the same enterprise customers.
"I think Anthropic is worth $3 trillion today," Gavin Baker, managing partner and chief investment officer of Atreides Management, said on the All-In Podcast in late June. "They are going to end this year with well over $100 billion." He added that annualized revenue could potentially reach "$200 billion to $300 billion by 2028."
Baker's view represents the bull case at its most aggressive: that Anthropic is not a cyclical beneficiary of an AI buildout but the defining infrastructure layer of a new computing regime, and that pricing it like a mature software company misses the regime shift entirely. The company's own fundraising supports part of that thesis. Anthropic raised $95 billion in 2026 alone, including a $30 billion Series G in February and the $65 billion Series H in May, in a round that drew strategic infrastructure partners including Micron, Samsung and SK Hynix, plus roughly $15 billion of previously committed hyperscaler investment, including $5 billion from Amazon.
That capital stack is not abstract. Amazon, Microsoft, Alphabet and Meta are expected to spend nearly $700 billion combined in 2026 to fuel their AI build-outs — a figure that helps explain both how Anthropic's revenue could scale so fast and why the industry needs it to.
Cyclical Boom or Structural Shift: The Call That Decides the Trade
Here is the judgment the market has not fully priced: the revenue surge is a hybrid, and the two components behave differently. The cyclical leg is the infrastructure supercycle — enterprises rushing to lock in API capacity, hyperscalers committing hundreds of billions of dollars to data-center construction, and a one-time repricing of what organizations will pay for coding and workflow automation. Cyclical legs mean-revert. When the buildout slows, growth rates normalize, and multiples compress. There are three historical analogs worth holding: the Cisco of the late 1990s, which traded at well over 100 times earnings at the peak of the internet buildout before the multiple collapsed when infrastructure spending plateaued; the cloud migration wave of 2020–2021, which pulled forward years of enterprise IT spending and then saw growth rates decelerate sharply; and the smartphone app economy, where the first wave of platform monetization produced outsized returns for the layer owners but brutal competition for everyone else.
The structural leg is different, and it is stronger than the cyclical comparison suggests. Anthropic's revenue is not being driven primarily by one-time setup fees or consulting; it is recurring API consumption embedded in enterprise workflows, coding agents that sit inside development environments, and model access priced on capability rather than cost. The company has held its premium output pricing rather than reverting to cheaper anchors, which suggests pricing power — a structural trait. If enterprises treat Claude as core infrastructure rather than an experiment, the revenue floor rises permanently, and the "run rate versus trailing revenue" gap narrows because the run rate becomes the new baseline rather than a peak.
So the call is this: the next 12 to 18 months are cyclical — a capacity-constrained sprint that will moderate — but the terminal state is structural, because the software layer Anthropic occupies is being rewritten, not rented. The risk is timing: if the cyclical leg turns before the structural leg is proven, a public investor buying at 20 times run rate has little margin of safety.
The Second-Order Question the Market Is Not Asking
The first-order read of the $100 billion headline is simple: AI demand is stronger than expected, buy the AI names. The second-order question is more uncomfortable. If Anthropic can grow from $9 billion to $100 billion in annualized revenue in roughly 24 months, what does that imply for the companies selling it the picks and shovels — and are they priced for the same curve?
The transmission chain runs like this: Anthropic's revenue growth requires compute capacity, which flows to Nvidia and the memory makers — Micron, Samsung and SK Hynix are now Anthropic investors or partners — which flows to the hyperscalers building the data centers, which flows back into model capability, which drives more enterprise adoption, which drives more Anthropic revenue. That is a virtuous circle — until it is not. The second-order risk is that the circle becomes a reflexive loop: revenue growth justifies more capital expenditure, more capital expenditure creates more capacity, more capacity requires more revenue growth to absorb it, and the industry begins selling capacity to itself. With nearly $700 billion in combined 2026 capital-expenditure commitments from Amazon, Microsoft, Alphabet and Meta, the loop is already large enough that a slowdown in absorption would show up as margin pressure across the whole chain, not just at the model layer.
There is also a cross-asset implication. A $2 trillion IPO that prices successfully would validate the entire AI capex thesis and likely reflate the megacap technology complex. A debut that stumbles — either because the pricing is cut or because early trading shows weak demand — would force a repricing of the capital-expenditure cycle itself, because the public market would be saying, in effect, that the revenue curve cannot justify the infrastructure spend. That is the third-order outcome: the IPO is not just a liquidity event for Anthropic's early investors; it is a stress test for the AI buildout's financial logic.
The Counter-Thesis, Stated at Full Strength
The strongest case against buying the Anthropic story at these levels is not that growth will stop. It is that growth will continue but not fast enough. A company can keep growing at 50 percent a year and still be a terrible investment if it was priced for 200 percent. The bear case rests on three pillars, each with a named advocate or data point.
First, the multiple is being applied to the wrong denominator. As noted above, run rate is not trailing revenue, and public-market investors who were burned in 2021–2022 by growth stocks that failed to convert hype into GAAP earnings may demand proof of booked revenue before paying a 20 times multiple. Second, the competitive moat is thinner than the revenue lead suggests. OpenAI, Google DeepMind and a field of well-funded challengers are all chasing the same enterprise budgets, and model capability has shown a tendency to converge — which pushes the industry toward price competition, exactly the dynamic that destroys software multiples. Third, the regulatory and political risk is rising: Anthropic had to temporarily disable two of its most advanced models in June to comply with a government export-control directive, and the company's relationship with the administration has been strained, including a Pentagon blacklist earlier in the year after military-use discussions soured.
"We look forward to deepening our government collaboration," Anthropic said in a June blog post. "We're also grateful to our users for bearing with us through this disruption, and to the researchers and industry partners who worked alongside us to make Fable 5 and Mythos 5 available again."
The falsifying signal for the bull case is specific and observable: if Anthropic's annualized run rate fails to reach the $100 billion lower bound by the end of 2026, or if the company's 2028 projection is revised down from the $190 billion to $200 billion range it has shared with investors, the structural-shift thesis is materially damaged. A $2 trillion valuation cannot survive a growth deceleration that leaves the run rate below $100 billion entering 2027.
What to Watch: Base Case, Upside, Downside
Short term (the IPO window): The base case is a debut in the fourth quarter of 2026 at a valuation between $1.5 trillion and $2 trillion, with pricing that validates the current private-market marks. The upside case is a hotter-than-expected book that pushes the listing above $2 trillion, lifting the entire AI complex. The downside case is a delayed or downsized offering that forces a private-markets reset.
Medium term (fundamentals, 2027): The base case is annualized revenue in the $120 billion to $150 billion range, with margins improving as compute costs per token fall. The upside case is that enterprise adoption of coding agents accelerates faster than expected, pushing the run rate toward $200 billion. The downside case is that capacity additions outstrip demand absorption, forcing price cuts and compressing the multiple.
Long term (structural, 2028 and beyond): The base case is that Anthropic establishes itself as one of two or three enduring model-layer leaders with durable recurring revenue. The upside case is that it becomes the default enterprise AI layer and justifies a multi-trillion valuation over a decade. The downside case is that model capability commoditizes, pricing power erodes, and the company trades like a utility rather than a platform.
The signals to watch are concrete: the actual S-1 filing (which will convert run-rate claims into auditable GAAP revenue), the IPO pricing and first-day trading range, quarterly run-rate updates against the $100 billion year-end expectation, and the 2028 projection in the company's investor materials. Any revision down in that 2028 number is the single most important red flag.
The Bottom Line
Anthropic's sprint toward $100 billion in annualized revenue is real, and it is historic. But a $2 trillion IPO at 17 to 20 times that run rate is not a bet on where the company is — it is a bet that the current growth rate is a new structural baseline rather than a cyclical peak. The evidence for a regime shift is stronger than the typical infrastructure-boom comparison suggests, because the revenue is recurring, priced on capability, and embedded in enterprise workflows. The risk is that the market has already paid for the regime shift in full, leaving no margin of safety if the cyclical leg turns first.
The cleanest way to state the trade: this is the market pricing the deficit of compute demand, not a cyclical dip — and if that deficit closes faster than the bulls expect, the multiple closes with it.
Explore more exclusive insights at nextfin.ai.
