Anthropic is planning to go public in September or early October, targeting a valuation of $965 billion, backed by $47 billion in annualized revenue fueled by Claude Code. But cheaper Chinese models, a strained relationship with the White House, and community resistance to data centers — three fires are waiting to be put out on the roadshow. It wants to beat OpenAI to the finish line, but going public first means absorbing punches first.
$47B Revenue Backs a $965B Valuation — Does Anthropic Dare Go First?
Anthropic is targeting a September IPO at a $965 billion valuation. The bet rides on Claude Code's growth. The risk? Three questions no one can easily answer.
Per The Wall Street Journal, Anthropic has entered pre-IPO talks with investors, aiming for a window between September and early October. That would rank among the largest IPOs ever — roughly 6.52 trillion yuan at current rates, per IT Home.
The valuation story hinges on Claude Code. Anthropic disclosed in May that annualized revenue had surpassed $47 billion, about 317.5 billion yuan. The price-to-earnings ratio sits just above 20x, below the heated reception SpaceX drew on private markets. In July, OpenAI self-reported annualized revenue exceeding its Q2 total, but that revenue leans on ChatGPT subscriptions and API calls. After Claude Code made coding a primary battleground, Anthropic's growth curve is steeper. Will it hold? That's not for us to say — it's for the investors stepping up in the coming weeks.
Anthropic has also been finalizing new compute deals with SpaceX and Google, and Claude Code usage data stays strong. But the same report concedes: services suffered intermittent outages this year. Short sentence, heavy weight — supply-side stability hangs over the valuation.
Three Fires — How Does the Roadshow Put Them Out?
Investors are buying whether the momentum can survive the next round of challenges. Anthropic's answer is one line: users only want the strongest, everything else is secondary.
Three lines of questioning surround the roadshow: cheaper Chinese AI systems, the relationship with the White House, and community resistance to data center construction across the US. Each one hits the growth curve directly. In recent weeks of pre-IPO meetings, investors have pressed hard on these points, sources say.
On China, Anthropic sticks to one unified message: the gap is still there, and it's not small. Dario Amodei's answer is one sentence — most users want the most intelligent model at any given time. Under this framing, less capable alternatives pose limited threat. Sources add that Chinese AI systems still trail top-tier models by at least several months overall. This "premium-only" thesis downgrades price competition to a secondary concern.
The White House is harder to handle. Anthropic's relationship with the Trump administration has tightened to the point of legal action. IT Home documented in related coverage that a judge called the US government's effort to block Anthropic unsupported by facts and unconstitutional. The company must both soothe regulators and prove it's not a political target.
Data centers meet the community. Opposition to building AI data centers — massive compute facilities that power AI training and inference — is rising across the US. Anthropic's antidote is storytelling: pushing AI into healthcare and biology, arguing this can ease social backlash against AI. Sources say the company has shared this plan with some investors, but specific timelines haven't been announced.
Anthropic frames the three issues separately: China is a "capability gap," the White House is a "legal battle," and community resistance is a "healthcare narrative." All three playbooks dilute near-term impact. But these narratives only hold if Chinese AI capability remains contained, government litigation doesn't break through existing rulings, and data center projects can withstand local elections. If any of these goes wrong, the growth assumptions behind $47 billion in annualized revenue will need rewriting.
First Mover Isn't Always an Advantage — First to Expose Is
OpenAI is waiting; Anthropic wants to ring the bell first. Racing to go public in September ahead of OpenAI is a counterintuitive bet: listing first means taking hits first.
Three fires are burning for Anthropic right now:
- Cheaper Chinese AI models squeezing its pricing
- Tensions with the Trump administration
- Local opposition to data center construction across the US
OpenAI isn't cornered — it's waiting, expected to push its IPO to as late as next year. Anthropic wants September or early October. Whichever exits private markets first will stand in the crosshairs of retail investors and short sellers. The reference point is SpaceX: after listing earlier, its market cap briefly surpassed $2 trillion, and optimistic analysts even projected Anthropic's valuation could exceed that level — but SpaceX's stock surged initially then quickly faced selling pressure. Wild swings are the default on the path from private to public, not the exception.
Anthropic's real motive for rushing is converting Claude Code's momentum into pricing power. In May, the company said annualized revenue had already exceeded $47 billion earlier that month, and in recent weeks it signed new compute supply deals with SpaceX and Google, with usage data showing strong demand for Claude Code and other tools. Listing while the revenue curve is still climbing steeply means financing firepower at its peak. Once growth slows, the valuation model gets rewritten.
Our take: a September listing is net positive for Anthropic overall, provided Claude Code revenue growth doesn't get cut in half within 90 days of listing. If growth drops below half of May's level, first-mover advantage quickly flips into first-mover baggage. The variables lie precisely in the three issues Anthropic's executives are trying to downplay. Any of these amplified during the IPO roadshow could cause that $965 billion figure to get repriced in public markets.
SpaceX's Script Is Still Warm
First to list, first to bleed. SpaceX's valuation briefly crossed $2 trillion on private markets, then cooled. Anthropic is targeting a $965 billion valuation — its pricing and post-listing performance will be the yardstick for how investors value top AI developers.
That script is still warm. Global AI investment has already reached trillions of dollars, most of it poured into compute, and it all rests on one assumption: "demand will keep growing rapidly." If Anthropic breaks issue price on day one, the entire sector's valuation logic needs recalculating. Anthropic executives downplayed Chinese AI competition in meetings, arguing that "users always want the smartest model" and that Chinese systems still trail top-tier models by at least several months overall. But there's a precondition buried in that: users will pay a premium for "the smartest." Chinese models are cheaper — the price difference is right there. Once the capability gap narrows to an acceptable range, the assumption wobbles.Watch These 5 Things, Wait for September Pricing
The listing is a September event, but the pricing logic is already running. Watch the key variables that will be written into the S-1 draft — each one determines whether that $965 billion is delivered or discounted.
The first red flag is Claude Code's retention. That $47 billion annualized revenue figure from May is an "annualized" metric — monthly revenue multiplied by 12. It's not GAAP annual revenue (Generally Accepted Accounting Principles, meaning actual booked income). The company's services also experienced "intermittent outages" this year. Once the S-1 filing is public, cross-check that number against subscription retention and API call volume rather than repeating it in isolation. The original report offers no third-party independent estimates, only the official "usage data shows demand remains strong" line, so what's available now is waiting for the S-1 to land and verifying it yourself.
Next, retest the "cooling down China competition" claim. Anthropic executives told investors that Chinese models "still trail top-tier models by at least several months overall," but the same article also states that "popular, cheaper Chinese AI systems" were a key pressure point from investors during the IPO roadshow. The tension between these two statements is itself an observation point: you can pull the latest pricing comparison from compute trading markets or major API rate cards, then check third-party leaderboards (like LMSYS Chatbot Arena, SuperCLUE) for the capability gap curve among top models. The original report provides no specific retest data, so this falls under "watch and calculate" rather than "conclude now."
Third is data center resistance and government relations. The original report flagged two threads: local opposition to data center construction, and Anthropic's tensions with the Trump administration. Both currently sit at the "exists" level, with no details on specific project stalls or policy actions — there's no way for ordinary people to verify in advance. What's available is following Anthropic's legal developments and local council data center bill reporting, then comparing once concrete events emerge.
Fourth is the healthcare and biology storyline. Anthropic told investors that AI applications in healthcare and biology can "ease some negative public sentiment" — a common way to translate social risk into product line expansion. Readers can check product pages or company blogs for any testable healthcare features or partner hospital lists, but the original report clearly states "specific plans haven't been announced," so this item is observation, not verification, for now.
Our take: the IPO will likely price somewhat below that $965 billion private valuation — the roadshow, media coverage, and post-roadshow sentiment will grind it down layer by layer. Dropping below $500 billion isn't realistic either; Claude Code's cash flow is right there. The precondition is that Claude Code retention doesn't collapse. If retention drops in any single quarter, the entire pricing logic needs rewriting.
Watch SEC EDGAR for the S-1 filing, verify the $47 billion annualized revenue metric and subscription retention, and don't reshare single-point numbers.
Pull the latest major API rate cards and compare pricing between Anthropic and top Chinese models at equivalent tiers to verify whether "cheaper" is marketing or fact.
Subscribe to third-party model leaderboards (human blind tests or automated benchmarks) and track capability gap curves among top models; don't blindly trust vendor self-reports.
Follow local data center siting proposals and Anthropic legal announcements; wait for concrete events before judging whether "resistance" has filtered down to the project level.
Check OpenAI's listing timeline and compare both companies' S-1 drafts side by side to distinguish "narrative" from "disclosure" — only the latter has legal force.
This article is based on the original IT Home report (citing The Wall Street Journal), dated 2026-08-11. Vendor-published figures (benchmarks, reductions, etc.) are official disclosures and have not been independently retested by third parties unless otherwise noted.