Intel Hits 70% Yield on 18A Node as Microsoft Commits to EMIB Packaging
Intel reports 70% functional yields on its 18A process node while securing Microsoft as its first public customer for EMIB advanced packaging technology.
Silicon, foundries, GPUs — the substrate of everything.
156 stories · page 1 of 7
Everything else on this site runs on what happens here. The desk follows semiconductors end to end: process nodes and the slipping roadmaps behind them, advanced packaging and the thermal ceilings it keeps running into, memory supply and the pricing cycles that swing the whole industry, and the fabs being built or quietly delayed. We read export controls and subsidy programmes as supply-chain events, not just politics, because that is how they land on a bill of materials. Vendor benchmark numbers get reported as vendor numbers. The through-line is capacity: who can actually manufacture what, in what volume, and by when — which is usually the real constraint on everyone else's plans.
Intel reports 70% functional yields on its 18A process node while securing Microsoft as its first public customer for EMIB advanced packaging technology.
A 'war room' of tech leaders and a 180,000-signature petition push for the reinstatement of Nvidia's founder as the board explores external candidates.
Independent researchers have found strong statistical evidence that major Chinese AI models were trained using export-restricted Nvidia Blackwell hardware.
Nvidia searches for a permanent CEO following Jensen Huang's departure, with the board evaluating internal and external candidates.
Nvidia co-founder Jensen Huang has stepped down following admissions of a relationship with an alleged Chinese spy, leaving CFO Colette Kress to steer the chipmaker through a leadership vacuum.
Department of Commerce export licenses for Nvidia's H200 AI processors have been granted to Chinese server assemblers operating out of empty offices and linked to sanctioned military suppliers.
U.S. intelligence officials believe Nvidia CEO Jensen Huang was targeted in an intelligence operation involving a top executive at AI chip buyer Megaspeed.
By acquiring Lumentum, Marvell gains critical indium phosphide manufacturing and laser expertise, aiming to close the gap on Broadcom’s vertically integrated optical interconnect portfolio for AI data centers.
AMD has issued Q4 2026 revenue guidance of approximately $17 billion, positioning the company to potentially surpass Intel’s quarterly revenue for the first time. This shift is driven by massive growth in data center deployments and AI accelerator adoption.
Smugglers are exploiting export control loopholes by packaging banned high-bandwidth memory from South Korea with cheap processors to bypass customs restrictions.
A new architectural experiment has successfully compiled a 32-bit RISC-V CPU into the weights of a Qwen 3.5 4B model, allowing an LLM to execute binary code.
The integration of Mellanox’s InfiniBand and Ethernet technologies has transformed Nvidia from a GPU vendor into a dominant full-stack networking powerhouse.
PsiQuantum has unveiled an eight-port optical switch prototype that leverages quantum-grade silicon photonics to reduce signal loss in high-density AI data center interconnects.
Intel is reshuffling its silicon roadmap, prioritizing the high-performance 18A-P node for Nova Lake while diverting Panther Lake capacity to meet surging Xeon demand.
GitHub reports a sixfold year-on-year increase in AI-generated code, necessitating a substantial expansion of its datacenter infrastructure with millions of additional CPU cores and petabytes of storage. This rapid scaling highlights the immense computational and storage demands
With 189,000 semiconductor jobs arriving by 2030, domestic fab expansion hinges on rapid technician pipelines rather than doctoral talent alone.
Nvidia is executing a massive supply chain strategy, securing multi-billion-dollar capacity reservations through 2028 for substrates, co-packaged optics, and OSAT services to sustain its AI accelerator roadmap.
Intel is reportedly exploring expanded collaboration with AMD to bolster its foundry business and standardise x86 architectures against rising ARM competition.
A new partnership between Nvidia and PreussenElektra explores converting shuttered German nuclear facilities into high-density data centers to solve power and grid constraints.
The discovery of Nvidia's restricted Blackwell Ultra B300 GPUs within state-owned China Telecom infrastructure highlights the persistent challenges of chip smuggling and grey market distribution.
A shift toward processing monazite ore could dismantle China's dominance in rare earth elements, offering a radioactive but viable path to domestic chip manufacturing security.
TSMC is actively pressing lithography and etch equipment vendors to eliminate Chinese rare-earth components from their systems to protect fab operations against escalating geopolitical supply chain disruptions.
Google and Amazon are releasing open source tools like TorchTPU and TorchNeuron to help developers port PyTorch apps to alternative AI chips without rewriting code.
A new software-based approach to Nvidia GPU configuration allows users to modify power limits and clock speeds via direct register overrides, bypassing the risks of traditional firmware flashing.