Sources: CoreWeave is exploring the use of financial derivatives as a potential hedge against a future drop in memory and storage chip prices
THE SO WHAT
When an AI cloud provider starts using derivatives to hedge memory and storage prices, GPU risk is no longer the only macro variable in your compute bill. Infra buyers should expect more complex pricing structures—and ask vendors explicitly how they’re managing component price swings over multi-year commitments.
READ THE SOURCE
MORE FROM THE WIRE
Tech & InnovationElon Musk Could Still Face Criminal Charges for Trying to Bribe Voters in Wisconsin
The line between online influence and alleged election interference is getting tested in court — that raises the odds of sharper rules on how platforms, political actors, and high-profile accounts can mobilize users. If your product touches political speech, civic engagement, or large-scale user incentives, assume more scrutiny on intent, targeting, and data use.
Tech & InnovationSamsung confirms new foldables will be revealed July 22 with new Flex Titanium display tech
If Flex Titanium meaningfully improves durability, foldables move from enthusiast hardware toward default form factor for power users. App builders should assume more split-screen, multi-state usage — and design workflows that exploit it.
Tech & InnovationInside Samsung Display's secret labs, where it builds and stress-tests next-gen displays — and says it's ready for Apple to launch a foldable: 'We mean it, we love other companies to join this market'
Display vendors openly inviting Apple into foldables is a tell — they want a category-defining anchor tenant to justify years of capex. If your roadmap assumes flat slabs, start prototyping experiences that treat folding and rolling as first-class states.
Tech & InnovationSamsung has a 16TB PCIe 6.0 SSD coming soon with read/write speeds of 28.4GBps and 21.9GBps respectively — but you won't be able to use it anyway
Storage bandwidth is now outpacing what most server and desktop architectures can practically feed — 28.4 GB/s reads are only useful if your PCIe lanes, CPUs, and workloads are ready. Infra teams planning AI and analytics clusters should be co-designing storage, interconnect, and compute, not buying each in isolation.