Cerebras Hardware Business Declines in Sign of ‘Lumpy’ Demand
THE SO WHAT
Cerebras’ hardware revenue dip underscores how non-standard AI accelerators face lumpy, project-based demand compared to the GPU default. If you’re considering alternative chips, model your risk around ecosystem maturity and deal timing — not just raw performance.
READ THE SOURCE
MORE FROM THE WIRE
Deep & Emerging TechUS Navy's new "Trump class" nuclear-powered program could hit $275 billion — with the first ship alone costing an eye-watering $23.4 billion
A $275 billion, nuclear-powered surface program that strains already-delayed yards is a signal that defense capex is locking in for decades, even as timelines slip. Shipbuilders and suppliers can plan for durable demand, but anyone selling “fast” dual-use tech into this stack needs to price in slow integration and chronic capacity bottlenecks.
Deep & Emerging TechChip stocks are closing in on a bull market. Here’s why some analysts are nervous.
Backlogs at AI infra vendors are pushing chip names toward a bull market, but analysts are already questioning how much of that order book will actually ship. If you’re planning long-term around today’s pricing and lead times, build scenarios where some hyperscaler and AI startup demand evaporates before hardware lands.
Deep & Emerging TechBank of America Invests $250B in Critical Infrastructure
A $250 billion, 18‑month push into digital, energy, and transport infrastructure means cheap capital is hunting for shovel-ready projects right now. Operators with credible AI, grid, or connectivity angles should be packaging projects to fit bankable templates instead of waiting on slower public funding cycles.
Deep & Emerging Tech'It took less than five seconds': Volkswagen badges are being targeted by thieves for the radar tech behind them
Thieves ripping off VW badges to get at radar units in under five seconds is a reminder that high-value sensors are now physical theft targets, not just cyber ones. Any connected hardware you ship into the field needs a threat model that includes component resale value and rapid, low-skill extraction.