‘You’re a liar.’ Why the world’s biggest building boom has run into a wall in California

An estimated $98 billion of data-center projects were blocked or delayed in H2 2025, and HMC StratCap withdrew its Monterey Park application after investing $40M to acquire a 200,000 sq ft site, citing new local restrictions and an upcoming June ballot ban. Regulatory, environmental and local opposition — plus high land and power costs — are pushing AI-related data center builds and associated construction jobs out of California, reducing capacity under construction and shifting market share to other states (Texas highlighted as a likely leader).

Analysis

The market is re-pricing the geography of compute capacity: regulatory friction and grid constraints are creating a multi-year arbitrage between high-cost, high-regulation coastal markets and lower-cost inland/centralized hubs. That reallocation amplifies demand for long-haul fiber, substations, modular power, and turnkey prefabricated shells — a supply-chain move that benefits manufacturers of large transformers, modular cooling, and on-site generation more than local commercial landlords.

Second-order winners will be firms that remove the political or grid dependency from site selection (onsite fuel cells, short-lead modular builds, third-party power procurement), while second-order losers include contractors and service providers concentrated in high-regulation jurisdictions whose backlog and bidding power will shrink over a 12–36 month window. Corporate capex timing is the key friction: hyperscalers can shift projects across states within 6–18 months, so expect lumpy quarterly guidance revisions rather than gradual demand erosion.

Tail risks center on federal intervention or standardized permitting that could snap capacity back to incumbents; conversely, accelerating state-level restrictions could hard-stop projects inside certain jurisdictions and force a permanent shift of both construction jobs and long-term hosting revenue. Monitor three near-term catalysts: (1) hyperscaler quarterly capex language over the next two earnings cycles, (2) public utility interconnection queue clearances over 3–12 months, and (3) any federal/legal rulings that create a nationwide permitting floor — each capable of flipping regional share dynamics within a year.

Key Decisions for Investors

• Long BE (Bloom Energy) 6–12 months: buy BE shares or call spread (e.g., buy 12-month ITM call and sell 12-month higher strike call) to capture faster adoption of on-site generation and microgrid solutions by data center developers. Risk/reward: target +35–50% if adoption accelerates; downside -30% if grid upgrades or cheaper alternatives prevail.
• Pair trade — Long MSFT / Short CBRE, 3–9 months: overweight MSFT (buy shares or 6–9 month call spread) to play margin upside from deferred hyperscaler capex and higher cloud pricing power; hedge macro real estate exposure by shorting CBRE (shares or 3–9 month put) which is exposed to cyclical relocations and slower leasing in high-regulation markets. Risk/reward: asymmetric — MSFT upside from multiple expansion vs CBRE downside if transaction volumes fall, target net +10–20% with capped drawdown of similar magnitude.
• Long AMZN (infrastructure exposure) 6–12 months via buy-write/covered-call: AMZN benefits from reallocated buildouts and incremental AWS demand even if physical builds move geography. Use covered calls to collect premium against near-term volatility; reward modest (10–25%) with downside protection via collected premium.
• Event-driven long GOOGL 3–9 months into permit/earnings catalysts: buy GOOGL ahead of quarterly commentary on data-center siting and energy contracts (or buy call options) — a favorable announcement on power procurement or green energy PPAs would re-rate capital intensity assumptions. Risk/reward: high gamma around earnings — expect 15–30% move on positive policy/contract news, otherwise tech-sector beta applies.