Hyperscalers might regret embracing natural gas if new forecast proves correct
THE SO WHAT
If US natural gas prices triple while hyperscalers are locking AI data centers into gas-heavy power deals, energy becomes a major variable cost risk, not a background line item. Infra teams should be modeling power price volatility alongside GPU supply — and pushing for more diversified or hedged energy mixes.
READ THE SOURCE
MORE FROM THE WIRE
Deep & Emerging TechHundreds of fake Chrome VPN extensions impersonating NordVPN, Proton, and more caught hijacking your traffic
Browser extensions are now a first-class security perimeter—737 fake VPN plugins quietly proxying traffic is a supply-chain attack on your users, not just a Chrome Store hygiene issue. Lock down extension policies on corporate machines and treat consumer VPN brand recognition as a weak trust signal unless you control the install path.
Deep & Emerging TechScottish police expect ‘great deal of public opposition’ to planned Larbert datacentre
Data centers are becoming political infrastructure—police planning for “incursion” and public resistance means every new site now carries community and security risk, not just planning risk. If you depend on hyperscale capacity, start mapping local opposition and permitting timelines into your capacity planning instead of assuming the cloud will just be there.
Deep & Emerging Tech'This one just needs a script': Researchers find ultimate Windows kill switch which can disable antivirus with almost no user interaction
A near one-click AV kill switch on Windows means assume endpoint defenses can be neutralized quickly—your security posture has to lean harder on identity, network segmentation, and rapid patching. Verify that April’s patch is deployed everywhere and run a tabletop on how your detection and response behaves once AV is gone.
Deep & Emerging TechAI Boom Reshapes Commodity Markets
When an AI buildout starts moving uranium, gold, and critical minerals, your AI roadmap is now entangled with commodity cycles and geopolitics. Large-scale AI users should treat power, fuel, and mineral exposure as part of capacity planning—hedging and long-term offtake may matter as much as GPU reservations.