Anthropic has surged to capture a dominant portion of new enterprise AI budgets, accounting for over 73% of spending among firms acquiring AI tools for the first time, according to customer-level data referenced by Jordan Klein, Mizuho's TMT specialist.
Klein noted a rapid change in vendor preference: roughly 10 weeks earlier the split between Anthropic and OpenAI was about even, and as recently as early December 2025 OpenAI held a roughly 60/40 advantage. That shift toward Anthropic, Mizuho says, is "super bullish" for Amazon because the cloud giant executes most of Anthropic's inference processing.
In addition to Amazon, Klein identified a set of companies he expects to see upside as enterprise AI adoption expands. Marvell Technology, Astera Labs, Fabrinet and Credo Technology were singled out as potential beneficiaries in Mizuho's view.
Mizuho frames these developments as evidence that enterprise AI adoption is at a turning point. As usage grows and real value and return on investment materialize, Klein wrote that adoption is "inflecting higher." He emphasized Anthropic's strength with AI coding tools as a distinctive edge that supports higher engagement and monetization.
The bank also stressed that inference demand is picking up and will be a key driver of capital expenditure and investment by hyperscale cloud operators into 2027 and 2028. That investment wave, Mizuho argues, will support construction of new agentic systems and applications, including initiatives referenced as OpenClaw. The firm added that rising memory consumption and potential power limitations are material considerations as deployment scales up.
Klein further listed Microsoft, Oracle, Nvidia, AMD, Micron Technology, Seagate Technology and Western Digital as companies positioned to gain from OpenAI's strategic emphasis on enterprise and coding-focused solutions. He described this repositioning as sensible and favorable for important suppliers across the stack.
Context and implications
While the data highlights a pronounced preference shift among first-time buyers toward Anthropic, Mizuho's analysis places particular emphasis on the downstream effects - from increased cloud compute demand to heightened hardware and memory requirements. The firm foresees these trends translating into multi-year capex cycles by hyperscalers and associated supplier demand.