SoftBank writes off $3.7 billion after cancelling AI data center turbine orders
The write-off exposes how fragile gas-turbine supply planning has become for companies building AI infrastructure.
The write-off exposes how fragile gas-turbine supply planning has become for companies building AI infrastructure.
If the claims hold beyond AMD's own tests, memory-constrained servers could need far less installed DRAM, though the rollout timing is still unclear.
Autonomous systems run from shore could cut crew numbers sharply, but their value depends on being linked into one networked fleet.
Security fixes for hundreds of previously disclosed vulnerabilities arrive alongside a reworked File Explorer context menu and AutoPlay dialog.
Tesla and SpaceX's chip venture is pressing toolmakers now, though its memory partner and technology remain undisclosed.
For maintainers of smaller projects buried under low-quality AI bug reports, the shift may offer some relief, though Torvalds says most other areas are not there yet.
The debate over whether the AI race with China can be shared or must be contested now turns on how far Washington loosens its chip restrictions.
By inheriting existing orders rather than waiting in the turbine queue, the company gains power capacity years sooner than new buyers can expect.
The Falcon 9, which has led global launches for a decade, is on a path toward retirement as the company shifts toward data and AI.
The practice is expected to spread from budget handsets toward premium models if memory prices keep rising into 2027.
The two systems target regulated buyers that need data protected during processing, and agent workloads that are limited by memory.
Users report possible SSD wear from the External Events Utility, but the scale of the damage remains unconfirmed until AMD ships a fix.
The expanded collaboration pairs national AI infrastructure with inference software, agent security tools and faster semiconductor simulation.
Randomising when these devices check their charge could remove the pattern, but many IoT devices are rarely patched.
The sample is unpolished and would need more processing before chipmakers could use it, and the company says a 300mm wafer may be possible before 2030.
Pull requests become a running record that agents update on each push, with merges blocked until review findings are closed.
Su expects data center revenue to keep climbing while overall PC sales fall by double digits in 2027.
Any move into memory would be years away and would not quickly ease the AI-driven DRAM shortage.
The gain matters most for listeners who want uncompressed audio over a wireless link, though only some devices will support the top tier.
Target volume production is 2030, with commercial production planned for 2028 and a focus on connectivity, radar and edge chips.