Memory Stock Valuations Spark Debate Over ‘Supercycle’ Potential
THE SO WHAT
If DRAM and NAND are printing record profits on AI demand yet still trade at a discount to GPU names, the market is underpricing the real bottleneck in the stack. Operators building AI-heavy products should assume memory pricing and availability are the next constraint after accelerators — lock in supply and design for bandwidth efficiency now.
READ THE SOURCE
MORE FROM THE WIRE
Deep & Emerging TechRedwood Materials lays off 10% in restructuring to chase energy storage business
Battery recycling is getting reprioritized around grid-scale storage demand—Redwood is cutting 10% to lean harder into energy storage, not EVs. If your decarb or infra roadmap still treats storage as a sidecar to vehicles, you’re misreading where the next decade of margin will sit.
Deep & Emerging TechBlackstone’s AirTrunk Plans Its First Data Center-Backed Bond
Data centers are now bond collateral—AirTrunk’s A$500M asset-backed deal turns AI-era compute into a securitized yield product. For operators, this means your colo and power contracts are feeding into fixed-income models, not just cloud P&Ls, tightening scrutiny on utilization and long-term offtake.
Deep & Emerging TechASMPT Sees Second-Quarter Sales Forecast Beat Due to AI Demand
Backend chip gear is printing again — AI demand is now driving upside not just for GPU vendors but for the packaging and assembly layer. If you rely on advanced packaging capacity, assume longer lead times and less pricing power than your 2025 models baked in.
Deep & Emerging Tech#AAN26: Roche’s multiple sclerosis pill delays relapse, but liver toxicity could need monitoring
Fenebrutinib doubling relapse-free intervals versus Aubagio while introducing liver-toxicity monitoring is the new normal in pharma — efficacy gains traded against tighter safety ops. Providers and payers need to budget not just for the drug but for the monitoring infrastructure and workflows that come stapled to it.