Data Centers

• Major companies—Amazon, Google, Meta, Microsoft, Oracle, OpenAI, and xAI—signed a federal pledge to provide their own electricity for AI data centers, shielding U.S. consumers from grid overload. • The commitment responds to exploding energy needs from AI infrastructure, following Meta's Louisiana gas plant initiative and similar efforts nationwide. • Industry leaders aim to prevent rate hikes and blackouts while fueling the AI boom central to U.S. tech competitiveness.

• ARM unveiled its first proprietary AGI CPU for AI data centers, shifting from licensing designs to manufacturing its own silicon. • Early adopters include Meta, OpenAI, Cloudflare, and Cerebras, positioning ARM directly in the AI infrastructure market. • This move underscores the intensifying competition in AI hardware, where control over end-to-end production is becoming critical for major players.

Lawmakers say moratorium on construction would buy time to create strong, federal guardrails for AIAmid an unprecedented energy crisis and the rapid buildout of artificial intelligence infrastructure, progressive lawmakers have unveiled a new policy to place a moratorium on the construction of AI datacenters.The policy, announced by Bernie Sanders, an independent senator from Vermont, and Alexandria Ocasio-Cortez, a New York Democratic representative, on Wednesday morning, aims to ensure the AI boom protects the environment and communities, and benefits workers instead of harming them. A temporary ban, the lawmakers say, would give the US government time to create strong federal safeguards for AI, which is “affecting everything from our economy and wellbeing to our democracy, warfare and our kids’ education”. Continue reading...

• The United States hosts nearly 5,500 data centers and accounts for three-quarters of the world's computing power as of May 2025, providing a substantial advantage over China in the global AI race according to EpochAI research. • America's semiconductor superiority remains the primary AI advantage, with Nvidia chips significantly outperforming China's most powerful Huawei-produced chips in both processing power and bandwidth memory, and the performance gap will increase substantially in coming years. • ByteDance plans to invest $23 billion this year on AI infrastructure including chips, while US companies are coordinating with the White House to secure additional grid capacity for data centers to prevent consumer price hikes.

• The Trump administration has granted Silicon Valley-based nuclear companies expedited pathways to test experimental reactor designs, with the DOE creating a new program and assigning concierge teams to help companies navigate bureaucracy. • About a dozen advanced reactor companies are currently planning to participate, with the administration targeting some reactors to achieve "critical" status—a key milestone toward functioning power plants—by July 2026. • The DOE quietly overhauled safety rules for new reactors and shared the revised regulations with companies before making them public, addressing long-standing complaints that companies lack opportunities to experiment.

• AI is expected to drive data center power consumption up 175% by 2030, creating unprecedented grid shortages that have pushed electricity prices higher across the country and compelled tech companies to seek alternative power sources. • Google's deal to power a new Minnesota data center blends wind and solar with a 30 gigawatt-hour battery from Form Energy, while the U.S. is expected to have nearly 65 gigawatts of battery storage capacity by year-end. • Dozens of startups including Amperesand, DG Matrix, Heron Power, Camus, GridBeyond, and Texture are developing power conversion and grid management technologies to address the critical energy shortage.

Amazon Web Services began deploying AI chips from Groq alongside its Trainium2 processors in US data centers on March 13, 2026, to diversify inference capabilities. The move supports over 100,000 Inferentia chips already in use, aiming to cut AI training costs by 50% for customers. This hybrid strategy enhances AWS competitiveness against Nvidia amid chip shortages, boosting scalability for enterprise AI workloads. Expansion to additional regions is slated for April 2026.