Nvidia may be ahead of AMD and Intel with PCIe Gen6-supporting platform for client PCs due to its client agentic AI ambitions, and that roadmap has suppliers like Silicon Motion paying attention.

Vice president of client storage solutions at Silicon Motion warns that the retail SSD market has almost disappeared as NAND makers prioritize shipments of memory to AI data centers.

Silicon Motion's Nelson Duann discusses NAND supply crisis in the consumer SSD market and the future of consumer storage.

Qualcomm is in talks to buy RISC-V-based AI accelerator and CPU developer Tenstorrent for $8 billion - $10 billion.

Next-generation data center processors from AMD and Intel with 16 DDR5 memory channels are even bigger than today’s designs.

TSMC is exploring panel-level packaging and is working on its CoPoS technology, but the company's Kevin Zhang says wafer-level packaging technologies is considerably more advanced than panel-level packaging.

AMD acquires MEXT to get Predictive Memory Engine that offloads infrequently accessed data from DRAM to NAND storage.

Marvell shares its vision for optically connected data centers, connecting devices across hundreds of kilometers, and the company already has hardware to build them.

Wiwynn is among the first to demonstrate Nvidia SCADA server that promises to offer AI systems petabytes of ultra-fast storage thanks to GPU-accelerated storage acceleration.

Biwin signs a 24-months supply agreement with an unknown NAND maker to get memory worth $1.86 billion.

TSMC is executing the largest manufacturing expansion in semiconductor industry history that combines simultaneous multi-fab N2 ramps, AI-driven manufacturing optimizations, and massive CoWoS/SoIC packaging capacity expansion to meet increasing demand for AI accelerators.

Nvidia and SK hynix have inked a multi-year collaboration agreement under which the companies will co-develop next-generation memory technologies for Nvidia's upcoming platforms and SK hynix will supply them to Nvidia.

Sales of Silicon Motion’s SSD controllers are record high, but supply of NAND for client applications may get worse in 2027, the company tells us.

AMD’s Helios set to compete against Nvidia’s NVL72 VR200 rack-scale system later this year, but its UALink-over-Ethernet interconnection may affect performance in certain workloads before real UALink interconnects are deployed.

Frore’s LiquidJet Nexus promises to enable 10% more token generation on Blackwell Ultra when compared to existing liquid-cooling solutions.

Counterfeit memory modules with G.Skill and V-Color badges sold at Chinese marketplaces use identical PCBs and heat spreaders, are hard to identify.

Astera Labs has shown off the Scorpio X-Series 320-lane PCIe switch that promises to enable vendor-agnostic scale-up capability for AI infrastructure and disaggregated data center infrastructure.

Supermicro demonstrates upcoming servers based on AMD’s EPYC ‘Venice’ CPUs, MI450 accelerators, and Nvidia’s Vera Rubin-based solutions.

Silicon Motion announces its SM2524XT mainstream SSD controller that promises 14 GB/s read speed, up to 2.5 million random IOPS, and sustained random performance.

SpaceX warns that it may not secure enough AI hardware to achieve its orbital AI ambitions even with supply from TeraFab. Also, Intel and Tesla may leave the TeraFab project.