
Meta leaves RE100 clean energy pact after a decade as its gas buildout for AI outpaces renewables
THE SO WHAT
Meta exiting RE100 to build 10 natural gas plants for AI load is a clear tell: hyperscale AI growth is outpacing available renewables and forcing a fossil backstop. If you're planning large AI deployments, power sourcing is now a reputational and regulatory risk surface—not just a line item in your cloud bill.
READ THE SOURCE
MORE FROM THE WIRE
Deep & Emerging TechEnvironmental Orgs Say Space-Based Data Centers Could Violate Federal Law
Space-based data centers move AI compute from land-use fights to orbital and atmospheric risk — ozone depletion and reentry debris now become part of your infra footprint. If your roadmap assumes “just put it in orbit,” add regulatory, liability, and ESG counsel to that fantasy before it hits a board deck.
Deep & Emerging TechThis energy company sidestepped Iran war challenges by pushing into data centers
An oilfield services company pivoting into data center and AI infra revenue is a clean example of how “energy” is re-bundling around compute demand, not just hydrocarbons. If you’re in traditional energy or industrials, your next growth line might be upstream of GPUs rather than barrels.
Deep & Emerging TechWhy Intel, Micron and other major chip stocks are falling — even as the rest of tech holds up
Memory and CPU names selling off while broader tech holds underscores how fragile the AI hardware narrative is to China risk, Korea sentiment, and single-vendor expectations. Operators should not assume today’s GPU-centric stack or vendor mix is stable — diversify supply and design for substitution.
Deep & Emerging TechNvidia reveals deep dive into what makes Vera CPU tick — with the super-powered Olympus core promising huge performance increases and more
An 88-core custom Olympus CPU with strong SPEC 2026 numbers is Nvidia saying the AI stack is CPU + GPU, not GPU-only. If you're planning next-gen clusters, revisit assumptions about where to spend on CPU versus accelerator for training, inference, and data prep.