Alphabet and Apple on January 12 announced a multi­-year collaboration under which Apple’s next­-generation Foundation Models will be built on Google’s Gemini models and run on Google Cloud infrastructure. After an extensive evaluation, Apple determined that Google’s AI technology offered the strongest combination of model performance, privacy controls and scalability. Under the agreement, Apple Intelligence workloads will continue to execute on Apple devices and on Apple’s Private Cloud Compute environment, with all user data encrypted and adhering to Apple’s strict privacy standards. The collaboration comes as Apple accelerates its AI roadmap following delays in delivering key features such as an improved Siri. Financial analysts had flagged concerns that Apple’s shares were underperforming relative to the Magnificent Seven due to uncertainty around its AI strategy; in response to the partnership announcement, consensus EPS estimates for Alphabet’s current fiscal year rose by 18% over the past year to $10.58, while Apple’s EPS estimates have also trended higher since mid-2025.

Clearway Energy disclosed three long-term power purchase agreements with Alphabet’s Google in 2025 covering 1.17 gigawatts of carbon-free generation capacity across Missouri, Texas and West Virginia. The total value of the agreements exceeds $2.4 billion over contract terms of up to 20 years. Construction on more than 1 gigawatt of new solar and storage capacity is slated to begin in 2026, with the initial projects expected online in 2027 and 2028. These deals expand an existing partnership, bringing Clearway’s total contracted capacity with Google to 1.24 gigawatts. The transactions underscore Google’s ongoing commitment to match its data center power usage with renewable energy, and they represent one of the largest corporate clean energy agreements of 2025 outside of the tech sector.

Alphabet has joined other leading hyperscale customers—Microsoft, AWS and CoreWeave—in planning to deploy Nvidia’s new Rubin platform for AI inference workloads during the second half of 2026. Rubin pairs Nvidia’s latest Vera CPUs and Rubin GPUs with proprietary high-speed networking, enabling AI inference token costs that are up to 90% lower than the previous-generation Blackwell platform. Industry research indicates that 95% of enterprise AI pilots fail to achieve meaningful ROI, in part due to high per-token expenses. By integrating Rubin across its data centers, Google expects to reduce inference costs dramatically on services such as Gemini-powered search summaries and its cloud-hosted Vertex AI offerings. This deployment is poised to accelerate customer adoption of agentic AI solutions and could boost Google Cloud’s AI infrastructure revenue share in the highly competitive cloud market.