Anthropic pulls ahead of OpenAI as revenue race heats up

Anthropic has crossed $30.0B in annualized revenue, overtaking OpenAI according to Jefferies. The company added roughly $21.0B in net new annualized revenue in three months (~36% of the $58.0B added by Jefferies' entire public software coverage, ex-Microsoft, in 2025), signaling rapid commercial adoption and a significant shift in competitive dynamics within the AI sector.

Analysis

Anthropic’s rapid commercial scale will shift economic capture toward cloud and infra providers: marginal dollar of AI spend disproportionately flows to GPU-backed hosting, telemetry and managed services rather than application-layer vendors. That dynamic favors hyperscalers with vertically integrated datacenter stacks and pricing power on reserved capacity, and it increases short-term capex intensity as they accelerate GPU procurement — expect sequential AWS and GCP capacity announcements and elevated supply-chain spend over the next 3–12 months.

A second-order beneficiary is contractual leverage: a high-volume model buyer can compress per-inference pricing, forcing model vendors to compete on price or accept lower take-rates; this will accelerate consolidation among smaller model and tool vendors within 6–24 months. Conversely, regulatory and procurement scrutiny becomes a lever for incumbents’ competitors — governments and large enterprises will demand cross-provider portability and auditability, which raises implementation costs and creates a window for middleware providers (and for cloud firms that sell secure, auditable enclaves).

Tail risks are concentrated and quick-moving: a major safety incident, large-scale data leak, or a sudden GPU supply squeeze can reverse adoption and re-price counterparty risk within days to weeks. Over the 12–24 month horizon, watch for enterprise renewals, AI SLAs migrating from usage to outcome-based contracts, and antitrust inquiries that could force structural separation or change commercial terms between cloud backers and the AI firms they fund.

Key Decisions for Investors

• Long AMZN (12–18 months): overweight AWS exposure to capture incremental GPU/managed-AI dollar. Use a costed call spread (buy Jan-2027 calls, sell a higher strike) to limit premium risk; target upside +35–60% if cloud pricing remains favorable, downside limited to premium (~100% loss of premium).
• Long GOOGL (9–15 months): buy shares or LEAP calls to play GCP wallet share and enterprise AI tooling (higher margin mix). Expect outsized re-rating if GCP demonstrates sustained AI capacity growth; protect with a 6–9 month 5% OTM put if regulatory headlines spike.
• Relative-value pair: long AMZN / short MSFT (6–12 months), sized to neutralize beta — idea is to capture faster infrastructure demand capture at AMZN vs MSFT’s more licensure/partnership exposure. Target gross return asymmetry +20–40% with downside defined by pair rebalancing and stop at 12% pair adverse move.
• Event hedge: buy 3–6 month ATM puts on MSFT sized to 25–50% of position notional to protect against sudden regulatory or partner-friction shocks tied to AI partnerships — cost justified by asymmetric downside if governments intervene.