IBM unveils sub-1-nanometer chip architecture that stacks 100 billion transistors onto a fingernail-sized processor
It's a research milestone for now, but the company expects it to become an industry standard within a decade
OpenAI debuts Jalapeño, its first custom AI chip to cut ChatGPT costs and reduce Nvidia dependency
The Broadcom-built ASIC is optimized for AI inference
Google is spending billions to turn its TPU chips into a real challenger to Nvidia
Google backs $3.2 billion data center deal to get its chips into Anthropic's hands, mirroring Nvidia's playbook
Samsung's foundry is booming as TSMC struggles with demand – bags Nvidia, Tesla, and Qualcomm as new clients
Google and AMD could also get their next-generation chips manufactured by Samsung
Intel could launch 'Raptor Lake Next' in 2027 as DDR4 makes a comeback
Industry sources say Intel will keep its older platform in the market
The PCIe 5.0 tax: Intel's Z990 chipset for Nova Lake runs hotter and uses more power despite shrinking
The motherboard designs can handle faster I/O and more demanding hardware
Intel launches Xeon 6+ Clearwater Forest with 288 E-cores on 18A node, teases Diamond Rapids for 2027
Intel just shipped its most ambitious server chip yet, Diamond Rapids promises PCIe 6.0 and twice the memory bandwidth next year
Nvidia RTX Spark CPU is now official: "superchip" will power Windows laptops and desktops
Nvidia wants to make your Windows PC an agentic and creator powerhouse
Why it matters: Nvidia just announced what it calls the most efficient PC chip ever built. RTX Spark is a Grace Blackwell system on a chip, 70 billion transistors on TSMC 3nm, with a Blackwell RTX GPU, a 20-core Arm CPU built with MediaTek, and up to 128GB of unified memory. It is purpose-built for agents, and it runs full RTX gaming and creation on the same thin-and-light laptop. It is a genuinely impressive piece of silicon.
Moore's law is hitting a wall, so researchers are stacking silicon chip layers instead of shrinking them
University of Illinois team stacks three active silicon layers on a single chip, achieving 98-100% transistor yield
Forward-looking: For years, the chip industry has chased better performance by shrinking transistors and squeezing more of them onto a flat slice of silicon. That strategy is running into hard limits. A group at the University of Illinois thinks the next gains will come not from going smaller, but from going vertical.
SK Hynix is embedding cooling into HBM memory to stop AI chips from overheating
New iHBM with cooling structures is said to cut thermal resistance by over 30%
Huawei says it can build 1.4nm chips without ASML's most advanced machines
Cut off from advanced lithography tools, Huawei is betting on stacked chip architectures and new packaging methods
Researchers built a switch 1,000 times faster than today's AI chips, and it barely generates any heat
University of Tokyo's antiferromagnetic spintronic switch operates at picosecond speeds
Intel CEO promises partnership with Nvidia will produce "exciting new products"
The chipmakers are developing SoCs with x86 CPUs and RTX GPUs
Apple and Intel are reportedly working on a chip manufacturing deal
TSMC capacity is tightening under AI demand – and the US government has been quietly pushing Apple toward Intel for months
Apple's success is running into a familiar problem: not enough chips
A record quarter, a $100 billion buyback, and a CEO transition
Samsung's chip profit jumped nearly 50-fold in a single year, execs warn the shortage will get worse
Hyperscalers are absorbing every advanced memory chip Samsung can produce
An AI agent just designed a complete RISC-V CPU from scratch in 12 hours
More bad news for human engineers and white-collar workers
If Apple built a $299 "Neo" desktop PC, Windows would have a real problem
The MacBook Neo already rattled the PC industry. Now imagine what Apple could do next
Crystal ball: Let's be clear upfront: there is no Mac Neo desktop. No leak, no supply chain rumor, no analyst note. What follows is conjecture – but conjecture worth taking seriously, because the pieces are sitting right there on Apple's workbench.
Meta doubles down on custom AI chips with Broadcom deal through 2029
Gigawatt-scale rollout and 2nm chips signal a massive AI infrastructure push
New terahertz technique lets engineers see inside running processors in real time
Modified lab equipment can map transistor activity inside a live processor without touching it
This Wi-Fi receiver can work inside a nuclear reactor, keeping robots connected
Highly anticipated: A team in Japan has developed a Wi-Fi receiver capable of operating in the extreme radiation inside nuclear reactors, an advance that could help robotics teams safely decommission aging power plants. The receiver, developed at the Institute of Science Tokyo, demonstrated resilience under radiation doses roughly 1,000 times higher than what typical electronics can withstand.
Chip foundry market hit $320 billion in 2025, driven by AI demand
TSMC is cashing in on the AI boom like no one else
This chip keeps working at 700°C, surviving lava-like heat
A USC team demonstrates a memristor device that keeps working far beyond current thermal limits
TSMC's 3nm capacity is now reserved for its biggest customers only
Capacity crunch leaves much of the chip industry waiting in line
Broadcom sounds the alarm as chip demand is pushing TSMC to its limits
From wafers to lasers and PCBs, Broadcom highlights a widening shortage across the semiconductor ecosystem