Anthropic has just surpassed OpenAI in quarterly revenue for the first time, with revenue more than doubling — just three months ago, OpenAI was still the clear leader.

The year-over-year numbers look extreme: Anthropic +142%; OpenAI only 18%.

But if you've been tracking Anthropic's recent monthly data, the growth slope has clearly moderated.

What makes it even more puzzling: Bloomberg recently reported Anthropic's annualized ARR above $65B (projected). Meanwhile, Jefferies noted that just two weeks ago, third-party trackers like Yipit and TickerTrends were still showing $70B–$80B.

Chart: Anthropic's growth curve flattens as OpenAI catches up

Pressure Points:

• Open-source AI is eating into the frontier-model franchise. A growing share of inference workloads now runs on cheaper small models and open-source alternatives — and market share is shifting accordingly. 
• At the same time, enterprises are scaling their own deployments to compress inference costs
•  AI infrastructure regulation has surfaced as a midterm election issue, potentially slowing the buildout of gigawatt-scale data centers. Ultimately, utilization trends drive revenue trends.

The Inflection Point Seen in the Secondary Market:

The price users are willing to pay per unit of compute — and the amount of compute required — is falling faster than token consumption is rising.

That raises a critical question: Has the steepest stretch of the AI growth curve, fueled by coding tools, already passed? Is the industry transitioning from mythical growth to a more normalized phase?

What Comes After Coding — And What If Nothing Does?

If a new use case on the scale of coding emerges in the next six months — something that can run reliably in production, scaling into verticals like finance, law, consulting, medicine, research and enterprise operations — the AI bull case remains intact.

But if that doesn't materialize, and revenue growth across the model makers begins to decelerate in unison, then today's $65 billion-plus Anthropic figure may, in hindsight, be recognized as the first clear yellow flag.

That, in turn, would ripple through the entire AI capex chain: Nvidia, memory, optical, servers, data centers and foundries.

What to Watch

The revenue divergence between Anthropic and OpenAI is, at its core, a story of shifting cloud consumption and enterprise adoption. The big cloud providers — not the model makers — are the clear winners. 

As enterprises pivot away from costly frontier-model APIs in favor of open-source alternatives, self-deployment, and vector retrieval, the economics of AI are being rewritten.

Core Beneficiaries

$Microsoft(MSFT.US) OpenAI‘s primary partner. A slowdown in OpenAI’s revenue or mounting consumer-side pressure could expose Azure Copilot monetization and cloud profit expectations to valuation re-rating risk.

$Oracle(ORCL.US)  OpenAI's infrastructure compute anchor, deeply tied to data center buildouts including the Stargate project.

The "Multi-Cloud Arbitrageurs" Among CSPs

$Amazon(AMZN.US) Houses the industry's strongest in-house inference silicon (Trainium/Inferentia) and is deeply integrated with Anthropic. Best positioned to capture enterprise demand pivoting toward cost-efficient AI infrastructure.

$Alphabet(GOOGL.US) TPU delivers strong cost-performance, with cloud arbitrage opportunities emerging in inference and open-source/self-deployment ecosystems.

ASIC & Networking – The Cost-Cutters' Choice

$Broadcom(AVGO.US) Absolute leader in enterprise custom inference ASICs and Ethernet networking switching chips.

$Marvell Tech(MRVL.US) – Accelerator and data center interconnect specialist, riding the custom hardware wave among cloud providers.

AI Coding & Enterprise API Adoption – Key Plays

$Palantir Tech(PLTR.US) Deep integration with Anthropic's Claude for agentic deployments across government and enterprise.

$MongoDB(MDB.US)& $Snowflake(SNOW.US) : Enterprise vector databases and data foundations — agent adoption, especially in coding and complex workflows, drives data retrieval API call volumes.

$GitLab(GTLB.US)& $Datadog(DDOG.US) : Developer ecosystem and observability tools, directly catalyzed by AI coding automation.

$Elastic NV(ESTC.US) Vector search and fine-tuning workflows. Open-source AI proliferation boosts token consumption for enterprise search and log analytics.

Defensive Hedges (Short-Term Only)

$Direxion Semicon Bear 3X(SOXS.US)/$Proshares UltraPro Short QQQ ETF(SQQQ.US)  – Semiconductor and tech 3x inverse ETFs. Use tactically if clear signals of AI hardware spending cuts emerge, as a hedge against richly valued compute names.