NEURA and SECO will build robot compute modules in Europe
NEURA and SECO will build robot compute modules in Europe

A German company designs the robot. An Italian company builds the board that goes inside it. An American company designs the chip that goes on the board. That is the European Physical AI stack as it actually exists. On Monday, two thirds of it signed a…

The US paid for a peace deal in Nvidia chips
The US paid for a peace deal in Nvidia chips

When Armenia opened the biggest AI factory in its region last month, we filed it under compute diplomacy. The Wall Street Journal now reports what the compute actually bought. Robbie Whelan reported on Friday that US negotiators used the promise of acc…

TSMC fab equipment demand nearly doubles in six months — AI surge pushes 2026 CapEx toward $64B amid tool shortages

TSMC’s equipment requirements have nearly doubled in just eight months as AI demand drives an unprecedented fab expansion, yet its 2026 CapEx budget has risen by only around 15%.

The current state of Hybrid Bonding in 2026 — TSMC sits at 6 microns and the HBM delay that nobody expected
The current state of Hybrid Bonding in 2026 — TSMC sits at 6 microns and the HBM delay that nobody expected

Hybrid bonding, the copper-to-copper joining technique that replaces solder microbumps in 3D chip stacks, is in high-volume production on logic and freshly postponed on memory.

China’s EUV technology ‘at a similar stage to ASML in 2004,’ analyst claims — Beijing’s semiconductor industry remains well behind Western rivals
China’s EUV technology ‘at a similar stage to ASML in 2004,’ analyst claims — Beijing’s semiconductor industry remains well behind Western rivals

Despite rumors, there is no evidence that Chinese makers of lithography tools can produce immersion lithography scanners in quantity.

Hot Chips 2026: Samsung reveals a three-phase HBM roadmap that puts logic and compute inside memory — zHBM ultimately stacks DRAM directly on top of the processor
Hot Chips 2026: Samsung reveals a three-phase HBM roadmap that puts logic and compute inside memory — zHBM ultimately stacks DRAM directly on top of the processor

Samsung detailed a three-phase HBM roadmap at Hot Chips 2026 that progressively moves logic into the base die and ultimately stacks DRAM directly on the processor.

Google will limit how much memory Android apps can use from February 2027
Google will limit how much memory Android apps can use from February 2027

Google will start policing how much memory Android apps use, and it says the reason is the global shortage of RAM. From February 2027, an app that exceeds the new limits risks two penalties. Android can slow it down or shut it. The Play Store can bury …

Intel 14A defect density is dropping faster than the company expected — ‘we have not seen this performance since 22nm,’ says CFO

Intel says defect density of 14A process technology is declining rapidly as internal teams are already developing 14A-based products, while external clients are now wondering about capacity that Intel can provide them.

EPA faces lawsuit over claims it fast-tracked approval of toxic ‘data center chemicals’ — exposure to photoacid generators used for semiconductor manufacturing could result in ‘sudden death,’ also appear to be long-lasting PFAS
EPA faces lawsuit over claims it fast-tracked approval of toxic ‘data center chemicals’ — exposure to photoacid generators used for semiconductor manufacturing could result in ‘sudden death,’ also appear to be long-lasting PFAS

An environmental group said that two new chemicals approved by the EPA for semiconductor manufacturing do not come with enough safeguards, so they’re suing the agency for ‘turning the new chemical review process on its head.’

Hot Chips 2026: Nvidia presents Groq 3 LPX architecture and unveils its first third-party inference benchmark — LP30-based rack already in production, company says
Hot Chips 2026: Nvidia presents Groq 3 LPX architecture and unveils its first third-party inference benchmark — LP30-based rack already in production, company says

Igor Arsovski, now Nvidia’s VP of hardware, presented the Groq 3 LPX rack’s architecture and published the first third-party benchmark of the hardware.