Artificial intelligence infrastructure spending is accelerating as leading technology and content companies lock in long-term compute capacity through massive multi-year deals. Across chips, cloud, and data centers, the common strategy is to secure supply, talent, and capacity ahead of competitors, with significant implications for capital expenditure, pricing power, and future margins. For finance teams, these moves provide early insight into capex cycles, utilization trends, and which vendors may gain margin leverage as artificial intelligence demand scales.
OpenAI sits at the center of many of the largest commitments. Amazon is considering an investment of around 10 billion in OpenAI, while Disney will invest 1 billion and grant OpenAI a three-year license to use characters from Star Wars, Pixar, and Marvel in Sora’s video generator and ChatGPT Images, excluding talent likeness and voices. Broadcom will co-develop OpenAI’s first in-house artificial intelligence processors, and AMD has signed a multi-year chip supply deal with OpenAI that includes an option for OpenAI to acquire up to roughly 10% of AMD. Nvidia plans to invest up to 100 billion in OpenAI and supply data-center chips, while Oracle reportedly agreed one of the biggest cloud deals ever with OpenAI to buy about 300 billion in compute over roughly five years, alongside a five-year, 11.9 billion contract between CoreWeave and OpenAI and the Stargate joint venture with up to 500 billion earmarked for artificial intelligence data centers.
Meta is also making an aggressive compute grab, agreeing to acquire Chinese startup Manus in a deal reportedly valuing Manus at 2-3 billion to help push agentic artificial intelligence across Facebook, Instagram, and WhatsApp. CoreWeave signed a 14 billion compute supply agreement with Meta, Oracle is in talks on a multi-year cloud deal worth about 20 billion, and Google agreed to a six-year cloud deal with Meta worth more than 10 billion, while Meta acquired a 49% stake in Scale AI for about 14.3 billion. Nvidia’s own footprint is expanding as CNBC reported it agreed to acquire Groq assets for 20 billion while licensing Groq chip technology and hiring its CEO and staff, joining Microsoft in investing up to 5 billion and up to 10 billion respectively in Anthropic as Anthropic commits 30 billion to run workloads on Microsoft’s cloud and up to 1 GW of compute with Nvidia’s Grace Blackwell and Vera Rubin hardware. An investor group including BlackRock, Microsoft, and Nvidia is buying Aligned Data Centers for 40 billion, Nvidia will invest 5 billion in Intel for roughly a 4% stake, and CoreWeave placed an initial 6.3 billion order with Nvidia with guarantees around unsold cloud capacity.
Google is simultaneously scaling its own capacity with a plan to invest 40 billion in three new Texas data centers through 2027 in Armstrong County and two in Haskell County, extending work on its Midlothian campus and Dallas cloud region within a 42-region global network, and it has hired key staff from code-generation startup Windsurf while agreeing to pay 2.4 billion in license fees for non-exclusive use of Windsurf technology. The article highlights how multi-year bookings and prepayments create deferred revenue and backlog visibility, while new chip supply paths and massive data center purchases intensify the race for power, land, and cooling. It warns of concentration risk as a small set of buyers such as OpenAI, Meta, and Anthropic pre-book capacity with a similarly concentrated supplier base including Nvidia, CoreWeave, Oracle, Microsoft, and Google. Finance teams are advised to monitor content licensing economics like Disney’s deal with OpenAI, the treatment of capitalized research and development, prepayments and multi-year revenue recognition, and to run 2026 scenarios where compute prices may fall 20-30% with new silicon or remain tight, reshaping total addressable markets, pricing, and margin dynamics across artificial intelligence-exposed vendors.
