Compute demand

Oracle refuted a Bloomberg report that its OpenAI data‑center rollout would slip to 2028, insisting all milestones remain on schedule despite labor and material shortages. The claim sent Oracle shares down over 4% after hours. The $300 billion, five‑year partnership with OpenAI aims to add 10 + gigawatts of AI compute, while Nvidia and Broadcom negotiate hardware deals. Supply‑chain constraints are pushing hyperscalers toward modular designs. Oracle sees the deal as a revenue‑diversification boost, offering integrated database‑cloud solutions and a high‑profile reference to attract more AI customers.

Google faces escalating AI service demand, requiring a doubling of serving capacity every six months. Google Cloud VP Amin Vahdat emphasized the critical need for AI infrastructure, revealing an ambitious goal of a 1000x increase in 4-5 years. CEO Sundar Pichai acknowledged an “intense” 2026 due to AI competition and addressed AI bubble concerns, highlighting Google’s strong cloud performance and disciplined investment. Capacity constraints limit deployment, exemplified by the Veo video tool. Executives underlined the drive for strategic efficiency alongside capital expenditure, emphasizing innovation and resource optimization.