Anthropic backers eye $2 trillion valuation. Its projected Q2 revenue was $10.9B

Investors have set their sights high for Anthropic, with some eyeing a valuation of $2 trillion or even $3 trillion. While the figures comes from investor models rather than a valuation target set by Anthropic, according to the Financial Times and CNBC, Anthropic could rival or best SpaceX’s record IPO, which raised $75 billion at an initial valuation of roughly $1.75 trillion in June.

What a $2 trillion valuation would require

The argument for Anthropic’s multi-trillion valuation is not new. Gavin Baker, managing partner and chief investment officer of Atreides Management, LP, said on the June 27 episode of the All-In Podcast, “I think Anthropic is worth $3 trillion today.” “They are going to end this year with well over $100 billion,” he continued. Baker said annualized revenue could potentially reach $200 billion to $300 billion by 2028. He argued Anthropic could become “very profitable” at that scale as its business becomes increasingly dominated by inference, citing reports of gross margins as high as 85% on inference workloads. On May 1, SemiAnalysis estimated that Anthropic’s gross margin on its inference infrastructure had risen from 38% to more than 70%, helped by falling token-production costs. SemiAnalysis later estimated that Opus 4.8 had inference gross margins above 85%.

While token costs may be falling, AI remains a “very capital-intensive business,” as Anthropic President Daniela Amodei told Bloomberg in early June. Training frontier models requires what she described as a “really big” upfront cost. To put that in perspective, Anthropic has already committed more than $130 billion to AWS and Microsoft Azure compute alone, including more than $100 billion to AWS over 10 years and $30 billion to Azure. It has also signed a multi-gigawatt agreement with Google and Broadcom whose value has not been disclosed.

How Anthropic compares with landmark IPOs

The table below compares Anthropic’s prospective public-market debut with several prominent IPOs, using reported offering valuations, the latest available pre-IPO revenue and profitability, and implied valuation-to-revenue multiples. Anthropic’s figures remain investor expectations rather than finalized offering terms.

Company	IPO	Valuation	Revenue basis	Profitability	Valuation / revenue	Sources
Amazon	May 1997	$438M	$15.746M (1996)	$5.777M net loss	27.8x	Amazon IPO announcement; Amazon financial results
Uber	May 2019	$82.4B	$11.27B (2018)	$3.03B operating loss; $997M GAAP net income	7.3x	Reuters; SEC prospectus
Facebook	May 2012	$104B	$3.71B (2011)	$1.00B net income	28.0x	Reuters; SEC filing
Alibaba	Sept. 2014	$167.6B	$8.463B (FY2014)	$3.76B net income attributable to Alibaba	19.8x	Alibaba IPO announcement; SEC prospectus
Saudi Aramco	Dec. 2019	$1.71T	$355.94B (2018 revenue and other income related to sales)	$111.07B net income	4.8x	Reuters; Saudi Aramco financial statements
SpaceX	June 2026	$1.77T	$18.67B (2025)	$4.94B net loss	94.8x	Reuters: IPO pricing; Reuters: financials
Anthropic (prospective)	IPO filed confidentially June 2026; investors expect an October listing	$2T+ investor expectation	Greater than $47B run rate in May 2026; investors project $100B–$120B annualized by year-end	Q2 2026E adjusted operating profit of $559M on $10.9B revenue, or about a 5.1% margin, excluding stock-based compensation	Less than 42.6x May run rate; 16.7x–20x projected year-end run rate at a $2T valuation	Reuters: IPO filing; Anthropic: May revenue run rate; Reuters: Q2 financial estimates; Financial Times: $2T valuation and year-end revenue expectations

Revenue growth is not the same as durable profit

At present, while AI revenues are quickly growing, relatively few companies within the AI sector have managed to reach operating profit, thanks to the technology. Exceptions remain a relatively small group of companies, especially those supplying the infrastructure behind it. NVIDIA, Broadcom, AMD, Micron Technology, TSMC, Supermicro and contract manufacturer Hon Hai (Foxconn) are among the clearest beneficiaries.  while many leading model developers and AI cloud providers continue to incur substantial losses.

At $2 trillion, Anthropic would be valued at about 42.6 times the $47 billion revenue run rate it disclosed in May, or 16.7 times the $120 billion year-end run rate its investors reportedly model. Neither run rate represents full-year revenue. Using the reported $4.8 billion of first-quarter revenue and projected $10.9 billion in the second quarter, followed by a gradual rise to a $10 billion December month, produces a 2026 revenue estimate in the ballpark of $60 billion. On that basis, $2 trillion is about 33 times the year’s sales. The eventual S-1 should also clarify whether sales routed through cloud partners should be compared on a gross or net basis.

Profit presents a second hurdle. Anthropic projected $559 million in adjusted operating profit on $10.9 billion in second-quarter revenue, a margin of about 5.1%. That figure included model-training costs but excluded stock-based compensation. The announcement of an adjusted operating profit also came during a brief window of time where a phenomenon known as “tokenmaxxing” was popular in which some enterprise companies briefly prioritized spending on token volume with few cost controls. The experiment quickly became unpopular as enterprise companies ranging from Uber to Microsoft realized the cost was unsustainable, and instituted caps.

In any event, Anthropic might not remain profitable for the full year as compute spending rises, according to Reuters Breakingviews. At the same 5.1% margin, a $120 billion revenue run rate would generate roughly $6.2 billion in annualized operating profit, putting a $2 trillion valuation at about 325 times that amount. Even at a 30% operating margin, it would imply about 56 times operating profit. The valuation pencils only if Anthropic delivers both the revenue growth and substantial margin expansion.

There is no clean public-market comparable

Even if Anthropic hit those numbers, investors still need a point of reference. There is no publicly traded U.S. frontier-model developer. The Financial Times reports that Anthropic backers have looked to Palantir and Nebius, which have traded at around 55 times revenue. But neither company has Anthropic’s operating model, and their valuations reflect enthusiasm for the same AI investment cycle Anthropic would be expected to validate.

History offers limited guidance. Companies that went public without profits, such as Amazon in 1997 and Uber in 2019, did so at valuations of $438 million and $82 billion, respectively. Companies that debuted near Anthropic’s target arrived with higher earnings. Saudi Aramco’s $1.7 trillion listing in 2019 rested on $111 billion in annual net income, roughly 15 times earnings. The closest precedent is SpaceX, which priced its June IPO at roughly $1.75 trillion on $18.7 billion in 2025 revenue and a GAAP net loss of nearly $5 billion, about 96 times trailing sales. While SpaceX’s stock performance has been uneven in recent weeks, it offers a profitable Starlink segment with $4.4 billion in operating income.

Enterprise demand will test Anthropic’s pricing power

Without a reliable comparable, Anthropic’s valuation rests more heavily on how much pricing power it can retain as enterprise buyers gain alternatives. Ramp’s August AI Index found Anthropic leading paid business adoption in July, reaching 43.5% of eligible businesses in its dataset. But adoption growth slowed, while Fable 5 accounted for only 11.4% of model-attributed spending on Anthropic’s models. Fable still generated roughly three-quarters as much model-attributed spending as OpenAI’s recently-introduced GPT-5.6 Sol, but the pattern suggests enterprise buyers are sorting workloads by price and performance rather than automatically concentrating spending on the most capable model.

Use of model-serving platforms, which can provide cheaper open-weight and Chinese-developed models, rose from 4.5% of AI-spending businesses in Ramp’s dataset in January to 6.1% in July. Ramp found that the growth had not yet meaningfully reduced spending on Anthropic or OpenAI, especially among first-time buyers.

Government restrictions remain a valuation risk

Another wrinkle is that Anthropic has recently had something of a rocky relationship with the U.S. government, which has shown it can intervene in how and where its models are distributed. Earlier this year, the Pentagon designated Anthropic a supply-chain risk after the company refused to remove restrictions involving autonomous weapons and surveillance of Americans. The underlying litigation remains unresolved. A federal judge temporarily blocked the government’s broader measures in March and said at a July hearing that the evidentiary record appeared to have grown worse for the government. But the designation and associated procurement risk have not disappeared.

Anthropic’s cutting-edge Fable 5 yielded a second confrontation. The Commerce Department imposed export controls on Fable and Mythos in June, forcing Anthropic to suspend them globally because it could not immediately verify every user’s nationality. The restrictions were lifted less than three weeks later after Anthropic and the government agreed to additional safeguards and monitoring.

Anthropic’s capital and compute commitments are intertwined

Despite some headwinds, Anthropic has frequently has emerged as a pace-setter in the genAI race, and has already raised $95 billion this year, including a $30 billion Series G and a $65 billion Series H in May. For perspective, Alphabet spent $61.1 billion on R&D in all of 2025, Microsoft spent $35.6 billion in its latest fiscal year and Apple spent $34.6 billion. Anthropic’s two 2026 financing rounds alone roughly equal the combined annual R&D budgets of Alphabet and Microsoft. The figures come from Alphabet’s 2025 Form 10-K, Microsoft’s fiscal 2026 Form 10-K and Apple’s 2025 Form 10-K.

Hyperscalers’ own spending is also expanding, with purchase commitments reaching nearly $1.5 trillion, according to the Financial Times.

One complication in gauging Anthropic’s potential valuation is the circularity of spending in the AI market. For instance, Amazon had already invested $8 billion in Anthropic. In April, it invested another $5 billion, with another $20 billion available subject to commercial milestones, putting Amazon’s potential cumulative investment at $33 billion. At the same time, Anthropic committed to spending more than $100 billion with AWS over 10 years, securing as much as 5 GW of capacity. AWS remains Anthropic’s primary cloud provider and training partner.