Enterprise AI spending has entered a period of cooling just as Anthropic, the developer of the Claude model family, prepares for one of the most anticipated public listings of 2026. While adoption rates for large language models remain high, new data suggests that a combination of aggressive price compression and enterprise optimization is curbing the explosive revenue growth that characterized the sector earlier this year.
Anthropic publicly filed its S-1 on October 26, 2026, targeting a Nasdaq listing under the ticker “ANT” later this month. The filing revealed that while the company reached $16.96 billion in Q3 revenue, its sequential growth slowed significantly to 47%, a sharp deceleration from the 144% growth recorded in the previous quarter. This slowdown coincides with broader market data from Ramp, which indicates that median per-employee AI spending at top firms has fallen nearly 10% to $7,205.
The divergence between usage and revenue is being driven by what analysts call “token price compression.” According to the Ramp AI Index, the effective price per million AI tokens has plummeted 41% in just eight months, dropping from a peak of $1.15 in March 2026 to $0.68 by August. This price war, coupled with enterprises “trading down” from flagship models like Claude Opus to more cost-effective mid-tier models like Claude Sonnet for routine tasks, has created a revenue plateau despite expanding user bases.
Anthropic has committed approximately $895.4 billion to future spending, primarily for computing infrastructure and data centers. These obligations dwarf the company’s current $132 billion cash pile, making the success of the IPO critical to its infrastructure roadmap.
The timing of the listing has also been complicated by a broader tech sell-off in early November 2026. Reports indicate that potential investors are now pushing for an IPO valuation closer to $1.5 trillion, a significant reduction from the $2 trillion targets initially discussed during the summer. This valuation pressure reflects growing caution over the “infrastructure-revenue gap,” as the cost of the compute required to train next-generation models remains high while the per-unit price of the resulting output continues to fall.
As Ramp data indicate that enterprise AI spending has stalled as of late 2026, the industry is shifting from a phase of “growth at any cost” to one of disciplined efficiency. For Anthropic, the “ANT” IPO will serve as a definitive test of whether a market leader can maintain its valuation in an era of thinning margins and maturing enterprise demand.