The Leftist Split Over AI Doom

A widening split on the US left over AI policy is shaping the regulatory debate between near-term harms—including jobs, surveillance and data-center expansion—and longer-term existential-risk concerns. Senator Bernie Sanders has backed a data-center moratorium and broader federal AI safeguards, while some Democratic Socialists argue that frontier-risk rhetoric can benefit companies such as OpenAI and Anthropic ahead of potential IPOs. A Politico poll found that two-thirds of Americans see at least some risk that AI could destroy humanity, increasing political pressure for comprehensive AI regulation.

Analysis

The investable signal is not a near-term federal ban; it is a widening permitting and political-risk premium on AI infrastructure. Local opposition can delay grid interconnects and campus construction well before national legislation advances, raising carrying costs for hyperscalers and weakening the volume assumptions embedded in VRT, ETN, GEV and data-center REIT valuations. Conversely, constrained power availability increases the strategic value of dispatchable, contracted generation and transmission capacity, favoring CEG and select regulated utilities with approved capex pipelines.

Over the next 1-3 months, policy rhetoric is more likely to affect high-duration AI beneficiaries than current earnings, especially names priced on 2027-28 data-center buildout. The key second-order effect is that slower physical deployment may redirect hyperscaler spend from new campuses toward utilization optimization, custom silicon, networking and power-procurement contracts; this is relatively supportive of AVGO and ANET versus construction-linked infrastructure vendors. A broad restriction on frontier models remains low-probability absent bipartisan legislative text, but state-level moratoria, water restrictions and interconnection queues can cumulatively produce the same capex-delay outcome.

Consensus appears too binary: either AI regulation destroys the buildout or it validates incumbents. A credible federal framework could actually entrench scaled labs and hyperscalers by imposing compliance costs that smaller model developers cannot absorb, while permitting constraints remain the more material bottleneck for physical infrastructure. DJT has no identifiable earnings sensitivity to this policy debate; avoid treating political association as a tradable linkage without a disclosed AI, data-center, or regulatory-business exposure.

Key Decisions for Investors

• Initiate a 3-6 month relative-value position: long CEG / short VRT, sized beta-neutral. The thesis is that power scarcity preserves CEG's pricing and contracting leverage while any campus-permitting delays hit VRT's order timing and high embedded growth expectations first. Falsify if VRT reports sustained backlog conversion and raises 2027 revenue targets, or if CEG's contracted-power pricing resets materially lower.
• Reduce unhedged exposure to data-center buildout proxies ETN and GEV into policy-driven strength; retain only where backlog is supported by transmission, utility and industrial demand rather than hyperscaler projects. Reassess after the next earnings cycle for disclosed data-center order concentration, cancellations and lead-time trends.
• For 6-18 months, prefer AVGO and ANET over physical-buildout suppliers as a less politically exposed AI-capex expression. Entry should follow evidence that hyperscalers are reallocating toward compute efficiency and network utilization; invalidate if cloud capex guidance falls broadly rather than merely shifting mix.
• Set an alert for state or municipal actions that suspend data-center permits, impose water/power conditions, or lengthen interconnection timelines. Two or more major-market actions within a quarter would justify increasing the CEG/VRT relative-value exposure; absence of such actions and continued hyperscaler campus approvals argues against the trade.