CLNM Capital 📨 U.S. Grid Crisis, EU 2026 Rules & the Real Cost of AI's Energy Appetite

Hello AI capital navigators!

This week reveals AI infrastructure's critical pressure points: Santa Clara data centers sit empty until 2028 awaiting power upgrades, while the EU prepares sweeping 2026 efficiency mandates. Tech giants rush to secure carbon credits as 45% of fund managers flag AI bubble risks amid a four-day market selloff. Meanwhile, Anthropic disrupted the first fully AI-orchestrated cyberattack, and Jeff Bezos returns as co-CEO of a $6.2B AI manufacturing startup.

Two major data centers in Santa Clara, home to Nvidia, remain vacant for years due to insufficient electricity from the local utility, highlighting a growing national bottleneck. Surging AI and cloud demand is overwhelming aging U.S. power infrastructure, delaying projects by years in high-demand regions. Despite massive investments and strong tenant interest, developers like Digital Realty and Stack must wait until at least 2028 for adequate power upgrades. 🔗 Bloomberg

Data centres now consume about 1.5% of global electricity and could more than double demand by 2030, driven largely by AI. The EU faces rising energy, water and emissions impacts, plus grid-connection constraints. In response, it is developing a 2026 energy-efficiency package, a new sustainability label, and a digitalisation roadmap. Recent legislation also mandates reporting to improve transparency and guide greener data-centre growth. 🔗 European Comission

VC funding for carbon and emissions tech rebounded to $3.5B in Q3 2025, driven by a few large late-stage deals, even as overall deal count hit a multi-year low. Western markets gained share, and AI-linked decarbonization technologies drew major investments. At the same time, tech giants like Microsoft, Google, and Amazon sharply increased purchases of costly, durable carbon-removal credits to offset rising data-center impacts. Microsoft alone has bought over 30 million credits — 20× more than any other buyer. 🔗 Reuters 🔗Q3 Carbon & Emissions Tech VC Trends report.

Microsoft will put in up to $5B and Nvidia up to $10B, while Anthropic commits to buying $30B in compute capacity from Microsoft. This deepens the interdependence between AI model provider, cloud infrastructure, and hardware — a deal that underscores both the scale and risk of “circular” AI financing. 🔗 CNBC

U.S. markets fell for a fourth straight session, with AI-linked tech giants like Nvidia and chipmakers dragging the Nasdaq down 1.2%. Investor concerns over high valuations and heavy debt for AI infrastructure are rising, with 45% of fund managers flagging a potential AI bubble as a top market risk. Globally, tech-driven losses spread to Asia and Europe, signaling broader investor caution. 🔗 The Wall Street Journal

Amid surging AI-driven power shortages idling data centers, this week's early-stage opportunities spotlight scalable infra and energy optimization:

CLNM Capital had the pleasure of meeting Doprax CEO Hemen Showkati at Web Summit 2025 sparking a connection over shared visions for scalable AI infra. Loved their sharp Web Summit wrap-up capturing the event's vibe. 🔗Read it here

With AI’s voracious power demand this week’s standout growth-stage opportunities are focused on the bottleneck:

Anthropic detected and disrupted a Chinese state-sponsored espionage campaign that used Claude Code to autonomously execute cyberattacks against roughly thirty targets including tech companies, financial institutions, and government agencies. The attackers jailbroke Claude to perform reconnaissance, exploit vulnerabilities, harvest credentials, and exfiltrate data with 80-90% automation and minimal human supervision— marking the first documented large-scale cyberattack executed without substantial human intervention. 🔗Anthropic

Jeff Bezos is returning to operations as co-CEO of Project Prometheus, a stealth AI startup he's partly funding with $6.2B, making it one of the best-backed early-stage ventures ever. Teaming with ex-Google X leader Vik Bajaj, it focuses on AI for engineering/manufacturing in aerospace, autos, and beyond, with 100+ hires from OpenAI, DeepMind, and Meta. 🔗 The New York Times

Tsinghua University filed 4,986 AI patents (2005-2024), surpassing Harvard, MIT, Stanford and Princeton combined, fueled by state support and a massive STEM pipeline. Its alumni drive startups like DeepSeek and Sapient. Yet US institutions retain higher patent citation impact and produced 40 notable AI models vs China’s 15 in 2024. With China’s share of top-tier AI talent rising fast, the quantity gap is closing — quality edge remains contested.🔗 Bloomberg

📅 Next: Jan 13, 2026 | ⏰ 15:00 Lisbon (60 min) | 🌐 Fully online & free (registration required)

Explores AI’s impact on productivity, labor markets, competition, and economic research methods. 60-min sessions with top academics, policymakers, and industry experts; interactive Q&A + open discussion. Chaired by Prof. Anton Korinek. Livestreamed & recorded. 🔗 Register

That’s it for this week.

Until next time,

The CLNM Capital

Want to share this newsletter with a friend? Use this link: https://clnm.beehiiv.com/