Anthropic's planned initial public offering — which could be the first for a frontier AI model developer — is being repriced in real time. Prospective investors are now discussing a valuation of $1.2 trillion to $1.3 trillion, down from the $1.4 trillion to $1.5 trillion range communicated to bankers just days earlier, and far below the $2 trillion figures floated in early October, Bloomberg reported Friday according to Investing.com [1]. The Wall Street Journal reported Thursday that three major prospective investors said the offering would need to be priced lower, with one suggesting a valuation closer to $1 trillion [1].

The reassessment follows Thursday's 6.4% plunge in the Nasdaq 100, which wiped out roughly $2.3 trillion in market value in its largest single-day decline on record [1]. The Cboe Volatility Index climbed to its highest level since April 2025 [1].

Valuation trajectory: $2T → $1.5T → $1.3T

Anthropic's valuation expectations have retreated in three steps. Early October discussions reached $2 trillion. After an investor briefing on financial performance, the range fell to $1.4–1.5 trillion. Thursday's market rout prompted the latest downward revision to $1.2–1.3 trillion [1]. Bloomberg's reporting suggests investors and bankers are still working toward an offering rather than abandoning the timetable [1].

Anthropic declined to comment on the valuation discussions [1][3].

Financials: explosive revenue, accounting-driven losses, massive compute commitments

The company's amended S-1 filing reveals the scale of both the opportunity and the capital intensity [2][6]:

• Revenue: $17 billion in Q3 2026, up 12× year-over-year. Nine-month revenue: $33.24 billion. Annualized run rate: ~$68 billion.
• Net loss: $23.3 billion in Q3; $104 billion for the first nine months. The filing shows 90% of the first-half loss ($72.4 billion of $80.5 billion) came from fair-value remeasurement of convertible notes — primarily from Amazon — whose carrying value rises as Anthropic's valuation climbs. This accounting effect largely disappears after the IPO when notes convert to equity [2].
• Operating performance: Roughly break-even in Q3 ($17.5B operating expenses vs. $17B revenue). Q2 showed ~$974 million operating profit. Operating cash flow was positive $3.9 billion in the first half; reports suggest free cash flow turned positive in Q3. Gross margin ~85% [2].
• Compute commitments: $895.4 billion in purchase and lease agreements, "substantially all" for compute and infrastructure. ~$447.5 billion due by end of 2030; ~$69.8 billion due in the 12 months after September 30, 2026 [2][6]. That near-term obligation is roughly four times current quarterly revenue.

The commitments explain why Anthropic needs to raise capital at scale. The IPO, led by Goldman Sachs and Morgan Stanley with 20 other banks, aims to raise at least as much as SpaceX's record $86.2 billion debut. The stock would trade on Nasdaq under ticker ANT [2].

METR finds 2× faster model improvement, driven by AI-assisted research

An independent evaluation by Model Evaluation and Threat Research (METR), given "unprecedented access to internal data" over roughly 11 weeks (mid-August through late October), estimated Anthropic is improving model capabilities on "well-specified tasks" around twice as fast as a year ago [4]. METR said it was able to "confidently" rule out compute growth as the driver, and said its guess was that the acceleration came from AI systems helping develop newer AI systems — the "closing the loop" dynamic Anthropic describes in its filing [4][6].

Caveats are significant. METR said the acceleration might be limited to tasks that can be graded automatically, not "messy" real-world tasks. It could also reflect understatement of prior progress. Internal disagreements and uncertainty among METR's team were noted [4]. Anthropic did not respond to requests for comment on the report [4].

CEO Dario Amodei publicly called for the industry to "slow the pace at which we improve the capabilities of AI models" in a September 12 essay, warning that recursive self-improvement could "outrun our ability to understand and control these systems" [4]. Amodei championed METR as an appropriate third-party evaluator; OpenAI's Sam Altman also supported the idea [4].

IPO timing uncertain: roadshow, October revenue, political headwinds

Reuters reported the roadshow was expected to kick off the week of November 9, with pricing the following week [3]. But four people familiar with investor discussions said potential investors are raising doubts about whether Anthropic can proceed on that timeline [3]. Investors are pushing for a more attractive valuation and are eagerly awaiting Anthropic's October revenue figure, which the company left blank in its updated S-1 [3].

A delay would amplify concerns about the AI ecosystem's health, particularly as ballot measures in multiple states limiting data center development have crystallized political opposition to AI infrastructure [3]. Microsoft tightened oversight of some data center projects from neocloud providers Nebius, IREN, and Nscale, Bloomberg reported Thursday; Nebius, Iren, and CoreWeave have each lost roughly half their value over the past month [3]. CoreWeave and SpaceX credit default swaps hit records implying rising default expectations [3].

Tech stocks have underperformed the S&P 500 by 10.7% over the past month — the most extreme one-month underperformance since May 2002, according to Truist Advisory Services [3]. Vanda Research said retail selling was the strongest since the COVID-19 pandemic [3].

What remains unknown

• Whether Anthropic proceeds at a $1.2–1.3T valuation, seeks a smaller raise, or delays past Thanksgiving.
• October revenue — a key data point investors are waiting for.
• How much of the $895.4B in commitments is firm versus contingent, and whether public-market investors will fund a capital plan that extends to 2045.
• Whether the "closing the loop" acceleration METR observed translates to broader capabilities or remains confined to benchmarkable tasks.

The distinction between a repricing and a postponement matters. If Anthropic goes public at a "more grounded valuation," it would test whether demand for leading AI businesses survives a sector correction [1]. For now, the company that says it wants to slow down is being forced to speed up its pitch to public markets.