This episode of the AI Daily Brief covers major developments in AI infrastructure, geopolitics, and competition. The discussion focuses on massive investment deals between tech giants and AI companies, the release of DeepSeek's V4 model, and escalating U.S.-China tensions in the AI space.

Key topics include Google's $40 billion investment in Anthropic, the White House invoking the Defense Production Act for grid infrastructure, and China's blocking of Meta's Manus acquisition. The episode connects these developments to broader themes around AI compute scarcity, energy constraints, and national security considerations in the global AI race.

Google's $40 Billion Anthropic Investment Signals Compute-for-Equity Trend

Google confirmed a $40 billion investment in Anthropic with $10 billion upfront and $30 billion contingent on undisclosed commercial milestones, following their previous $3 billion investment in 2023-2025.

The deal mirrors Amazon's structure of $5 billion upfront plus $20 billion contingent, with Anthropic committing to spend $100 billion with AWS over the coming decade.

Mirai Securities noted that each gigawatt of capacity equals a full-scale nuclear reactor, with Anthropic securing incremental capacity rivaling Microsoft's entire 6-gigawatt global data center footprint.

Google's ownership stake could exceed 20% if the full investment is made, up from their previously reported 14% stake as of March 2024.

White House Invokes Defense Production Act for Grid Infrastructure

President Trump declared grid infrastructure and supply chains essential to national defense, authorizing the Secretary of Energy to make purchases and commitments to expand domestic capabilities.

Goldman Sachs identified electricity as AI's next bottleneck, projecting data center demand will grow from 6% to 11% of U.S. electricity by 2030.

JPMorgan called the aging grid a national security risk, writing that grid resilience efforts 'increasingly underpin economic development, industrial competitiveness, and national security.'

The Financial Times noted that boosting U.S. power grid faces 'complex web of regulatory, financial, and supply chain challenges' with project backlogs creating major choke points.

DeepSeek V4 Delivers Competitive Performance at Fraction of Cost

DeepSeek V4 Pro performs comparably to Opus 4.5 and GPT-5.3 on benchmarks while costing 17¢ per million input tokens versus $1.20+ for frontier models.

Matthew Berman argued DeepSeek V4 represents a 'serious threat' because 'most use cases don't need absolute frontier intelligence' and the cost difference makes it attractive for business applications.

DeepSeek stated they are 'currently limited by compute supply and will drop prices even more once Huawei production is ramped up in the second half of the year.'

Former Trump AI advisor Dean Ball noted that while V4 'is further behind' than DeepSeek's earlier R1 model, it still represents significant value at its price point.

China Blocks Meta's Manus Deal Amid Rising AI Tensions

Beijing blocked Meta's $2 billion acquisition of Manus on national security grounds, requiring the parties to 'terminate and revoke the acquisition transaction.'

Chinese officials told the Financial Times the deal was viewed as 'a conspiratorial effort to drain China of AI talent and resources.'

China plans to curb U.S. investment in domestic tech companies, instructing firms like Moonshot and StepFun to reject U.S. capital unless explicitly approved.

Two Manus co-founders attended questioning in China in March and were told they could not leave the country while the investigation was ongoing.