Semiconductor stocks are surging while the ‘Magnificent Seven’ is struggling. This divergence of fortunes could be bad news for the market.
THE SO WHAT
AI capex is now visibly accruing to the picks-and-shovels layer while the demand side that’s funding it trades down—your risk is being overexposed to buyers and underexposed to suppliers. If you’re building in the AI stack, assume your customers will be more valuation- and cash-constrained than your upstream infra vendors and price, contract, and runway accordingly.
READ THE SOURCE
MORE FROM THE WIRE
Deep & Emerging TechJIJ Appoints Steven Gibson to Lead U.S. Expansion
Hybrid quantum–AI optimization is moving from lab curiosity to go-to-market motion—JIJ putting a President of Americas and a dedicated solution architect on the ground is a bet that U.S. enterprises will pay for combinatorial optimization before full-scale quantum arrives. If you run complex logistics, pricing, or scheduling, it’s time to at least map where quantum-inspired solvers could plug into existing AI workflows.
Deep & Emerging TechNIST Taps SRI to Establish The Quantum Manufacturing Engineering Center (QMEC)
A dedicated Quantum Manufacturing Engineering Center at SRI backed by NIST is a shift from ‘more qubits’ to ‘how do we build these at scale’—manufacturing, packaging, and reliability become the new bottlenecks. Hardware and materials teams should read this as a green light to pursue quantum-adjacent process IP and partnerships, not just algorithm work.
Deep & Emerging TechWhy quantum computing may be the White Houses new AI
If the White House starts talking about quantum the way it talks about AI, expect a policy and funding pivot toward quantum infrastructure, standards, and workforce. For operators, that means more public money and regulatory attention around cryptography, secure comms, and long-horizon R&D—worth tracking if you’re exposed to post-quantum risk or can tap federal programs.
Deep & Emerging Tech'Near zero-cost memory expansion through recycling': Meta will reuse terabytes worth of DDR4 memory using CXL tech and avoid paying the RAM tax
Reusing retired DDR4 via CXL to expand memory at near zero cost is a reminder that AI infra economics are now fought at the component and interconnect layer, not just GPUs. If you’re running large in-memory or AI workloads, push your vendors on CXL roadmaps and memory pooling—there’s real margin in sweating “obsolete” assets longer.