TechstrongSemi

The US Commerce Department plans to award up to $874 million to seven semiconductor companies, with the largest sum going to GlobalFoundries for co-packaged optics and others targeting advanced memory, packaging, and AI hardware. The awards extend the CHIPS Act's push into the research layer, with the government taking minority equity stakes in the recipients. Read more about the details of the seven projects and their strategic importance here: https://lnkd.in/gT6XnVJV #Semiconductors #AgenticAI #Research #CHIPSAct

GlobalFoundries has secured a $300 million CHIPS research award to accelerate silicon photonics and co-packaged optics, targeting the copper interconnects that bottleneck data movement inside AI clusters. The funding builds on the company's SCALE platform and previous manufacturing grants, positioning optical interconnects as a critical enabler for the bandwidth and energy efficiency demands of next-generation AI infrastructure. Read the full article for more on GlobalFoundries' silicon photonics strategy: https://buff.ly/eISaexG #CHIPSAct #SiliconPhotonics #AgenticAI #Semiconductors

Samsung and Broadcom have signed an MOU for a strategic semiconductor partnership expected to exceed $200 billion through 2030, spanning advanced manufacturing, high-bandwidth memory, and chip packaging. The deal bolsters Samsung’s foundry ambitions as Broadcom’s custom AI chip business expands, though Samsung still trails TSMC in the contract manufacturing race. Read the full article for more on the Samsung-Broadcom partnership: https://buff.ly/cGxp7HX #Samsung #Broadcom #AI #Semiconductors

Enterprise processor design continues to evolve as infrastructure demands change. Intel plans to bring Hyper-Threading back to future Xeon processors, reflecting a renewed focus on simultaneous multithreading for virtualization, databases, cloud infrastructure, and AI inference workloads. Read more: https://lnkd.in/g8S8ea4M #Intel #Xeon #DataCenters #EnterpriseIT

For most of the computer era, Moore's Law meant shrinking transistors sideways. IBM's nanostack just proved it can go vertical instead, stacking N-type and P-type layers to nearly double the density of its 2nm technology. The semiconductor roadmap that defined computing for decades didn't end. It found a new axis. Our latest looks at how building silicon upward changes the future of chip design and why the SRAM scaling buried in the announcement may matter most for AI: https://buff.ly/PihkCrK #IBM #Nanostack #MooresLaw #Semiconductors

Intel's data center business surged 59%, agentic AI workloads are driving demand for server CPUs, and hyperscalers are locking in multi-year deals. The foundry added Tesla and Fortinet, capex is heading to $20 billion, and the numbers are telling a different story for a company that was supposed to be left behind by GPUs. Read how Intel is riding the AI infrastructure wave: https://lnkd.in/gRfYY8dg #Intel #AI #DataCenter #Semiconductors

AMD is mounting its strongest challenge to NVIDIA yet, unveiling its Helios rack-scale AI system alongside a $5 billion investment in Anthropic that secures up to 2 gigawatts of Instinct GPU capacity. The Anthropic and Microsoft Azure deals are part of a broader push to capture inference workloads currently dominated by NVIDIA, as demand for agentic AI accelerates. Get the full story on how AMD is taking on NVIDIA in the data center arena: https://lnkd.in/gCVWSK5P #NVIDIA #Helios #AI #DataCenter

A new report projects the data center chip market will reach $1.2 trillion by 2030, driven by a shift from model training to agentic inference that pushes inference accelerators to 73% of revenue. High-bandwidth memory, custom XPUs, and a CPU resurgence are reshaping the supply chain, with hyperscalers remaining the largest buyers while tier-2 cloud providers grow fastest. Read the full forecast for the hardware trends defining the inference era: https://lnkd.in/gj7sNJM7 #AgenticAI #Semiconductors #DataCenters

For most of computing history, progress meant separating software from the machine. Google is now preparing to reverse that with an experimental chip that bakes parts of the Gemini architecture directly into silicon, betting the model is ready to become the machine. Our latest looks at the efficiency gains and the risk of freezing yesterday's architecture into tomorrow's hardware: https://buff.ly/Dfc6Pt1 #Google #Gemini #AgenticAI #Semiconductors

AI competition is increasingly shaped by infrastructure as much as software. Samsung is reportedly in talks to invest in Mistral AI, potentially strengthening its position in the AI hardware supply chain. The discussions highlight the growing importance of partnerships between chipmakers and AI developers. Read the full article to learn what the proposed deal could mean for the future of AI infrastructure: https://lnkd.in/gJRuDbUH #AIInfrastructure #Semiconductors #Samsung #MistralAI