AI Firms Commit Billions As Demand Drives Infrastructure Spending

Technology companies are committing billions of dollars to chips, cloud capacity and data centres as demand for artificial intelligence pushes firms to secure computing infrastructure.

The spending has created a web of partnerships between AI developers, chipmakers and cloud providers, with several agreements combining infrastructure purchases with strategic investments. Reuters reported that AMD, Nvidia, Google, Microsoft and Meta were among companies committing large sums to support AI workloads.

The latest agreement came on July 22, when Advanced Micro Devices agreed to sell Anthropic tens of billions of dollars worth of AI servers and invest up to $5 billion in the Claude maker, Reuters reported. Anthropic will deploy up to two gigawatts of AMD Instinct MI450-series GPUs, with the first gigawatt scheduled to begin deployment in the first half of 2027, AMD said.

The AMD-Anthropic agreement showed how chip suppliers were increasingly becoming financial partners of the AI companies that depend on their hardware. AMD said its investment would be tied to the achievement of deployment milestones, while the companies will also work together on optimising Claude workloads for AMD GPUs.

Nvidia has pursued a similar model with OpenAI. Nvidia announced in September 2025 that it intended to invest up to $100 billion in OpenAI progressively as the companies deploy at least 10 gigawatts of Nvidia systems for OpenAI's next-generation AI infrastructure. The first gigawatt was scheduled for deployment in the second half of 2026.

OpenAI has also secured infrastructure through other technology companies. Reuters reported that Oracle had agreed to one of the largest cloud computing arrangements involving OpenAI, with the AI company expected to purchase about $300 billion of computing power over roughly five years. CoreWeave separately signed a five-year contract worth $11.9 billion with OpenAI.

Amazon was considering an investment of about $10 billion in OpenAI, according to the Reuters report, while Disney agreed to invest $1 billion in OpenAI and entered a three-year licensing arrangement allowing selected Disney, Pixar and Marvel characters to be used in Sora.

Meta has also been securing large-scale computing capacity. The company agreed with AMD in February 2026 to deploy up to six gigawatts of AMD Instinct GPUs for its AI infrastructure, with initial GPU deployments scheduled for the second half of 2026. Meta said the agreement was part of a broader strategy to use multiple chip suppliers rather than depend on one architecture.

Meta's infrastructure strategy also includes internally developed chips. The company said in March that it was developing four new generations of its MTIA custom chips over two years, while continuing to source silicon from other industry suppliers.

Nvidia has expanded beyond selling GPUs by investing across the wider AI infrastructure chain. Reuters reported that Nvidia agreed to invest $2 billion each in photonic technology companies Lumentum and Coherent, while an investor group including Nvidia, Microsoft and BlackRock agreed to acquire Aligned Data Centers in a transaction valued at $40 billion.

Google is also increasing infrastructure spending. Reuters reported that Google planned to invest $40 billion in three new Texas data centres through 2027, adding capacity to its existing US cloud infrastructure. Alphabet also planned to invest up to $40 billion in Anthropic, deepening its relationship with the AI company.

The infrastructure race extends to specialised providers. Nebius agreed to provide Microsoft with GPU infrastructure capacity worth $17.4 billion over five years, while CoreWeave expanded its cloud arrangement with Meta to $21 billion, according to Reuters.

The scale of these commitments reflects the industry's expectation that AI workloads will require substantially more computing capacity. Companies are therefore securing chips, cloud capacity, data centres and specialised infrastructure years before some deployments are scheduled to begin.

The next phase will depend on whether planned data centres, chips and power capacity are deployed on schedule and whether demand for AI services continues to justify the capital being committed. AMD's first Anthropic deployment is scheduled for the first half of 2027, while Nvidia's first gigawatt under its OpenAI partnership is scheduled for the second half of 2026