US Midterms Deal a Blow to AI as Compute Expansion Hits a Wall
Key Takeaways
- US voters overwhelmingly backed restrictions on AI and data centers
- Texas shifts, and its status as a key hub for AI data center construction may be shaken
- OpenAI's fundraising runs cold as Anthropic gains the upper hand in the enterprise market
- Apple launches a developer trial of Apple Intelligence in China
Election Results Put the Brakes on AI Expansion
The US midterms are the most important variable for the AI sector today. According to Ars Technica, US voters overwhelmingly voted to restrict AI and data centers, while the tech industry frames the backlash as "anti-Trump" rather than "anti-tech" (source: Ars Technica). Whether that explanation holds will determine where policy goes next.
The immediate casualty is compute infrastructure. Texas is the fastest-growing data center market in the US and the most concentrated area of AI buildout by big tech, but Democrat James Talarico defeated Ken Paxton to win a Senate seat and Gina Hinojosa defeated the Republican incumbent governor, and the shifting political map could jeopardize Texas's data center standing (source: MIT Technology Review).
The tech giants' political spending failed to deliver the expected results. According to the FT, big tech's roughly $600 million in midterm campaign spending had little effect, Musk has come under fire, and discontent with AI and Silicon Valley helped lift Democratic support (source: FT Technology).
Capital Divergence: OpenAI Runs Cold, Anthropic Gains the Upper Hand
A clear split has emerged on the funding side. According to Newcomer, interest in OpenAI's latest funding round was lukewarm, while Anthropic's edge in the enterprise market has forced OpenAI into a hurried search for strategy (source: Newcomer). The contrast shows that enterprise deployment capability is becoming a key variable in the valuation narrative, though the hot-or-cold of a single funding round is not enough to determine the long-term landscape.
At the same time, agents as an interface are reshaping the cost structure. Chip Strat's newsletter discussed agent economics, who bears inference costs, vertical integration, and what this means for Google, Amazon, Microsoft, Meta, OpenAI, Anthropic, Apple and semiconductor makers (source: Chip Strat).
Marginal Changes in Products and Supply Chains
Apple launched a developer trial of its in-house AI system Apple Intelligence in China on Nov 4, about two years later than markets such as the US (source: kaopunews). The pace of rollout in the China market remains a weak spot in Apple's AI strategy.
Supply chains are also tilting toward AI. Macronix, one of Nintendo's ROM chip suppliers, is reportedly shutting down that business and reallocating capacity to the NOR and NAND flash chips needed by the AI industry (source: Tom's Hardware).
On the risk front, Bill Gates warned at the FT's Future of AI summit that AI could cause a pandemic "much worse than COVID" and said AI bioweapons "could be worse than COVID," calling for stronger government regulation (source: FT Technology).
Connecting today's threads: the constraints on AI are shifting from technology to politics and capital — voters, state-level politics and funding heat will change the pace of compute and products faster than model parameters.