Latest news, developments and reporting on AI Chips & Compute.
Seoul’s KOSPI closed up 2.05% at 5,668.05 on Oct. 16, snapping a seven-day losing streak, as chip shares recovered and Fed Chair Kevin Warsh’s comments on the U.S. economy and AI boosted risk appetite. The won rose to 1,331.4 per dollar.
Meta released Muse Spark Open, an 816B open-weight MoE with 42B active parameters, 1M context, multimodal input, and deployment on two Mac Studios.
SK hynix showcased next-generation AI memory technologies at OCP Global Summit 2026, including MARS, High Bandwidth Flash progress and a second-generation 256GB CXL 3.2 product.
Data center hardware startup Mueon has raised $65 million in a Series A round led by Maverick Silicon to develop Cubelet, a three-dimensional server architecture that stacks computing and memory components.
A survey by recruitment platform Incruit shows SK hynix was the most-followed company among South Korean job seekers over the past year, ahead of Hyundai Motor, Incheon International Airport, Kakao and Samsung Electronics.
The IMF warns that weaker AI investment, inflation and cyber threats could tighten financial conditions, while South Korea's record chip profits fail to prevent market panic and trading halts.
South Korean newspapers on Oct. 16 led with a KOSPI decline of more than 7 percent driven by chip stocks, citing doubts over artificial intelligence and fears about interest rates.
NVIDIA released Nemotron 3.5 Super VL, a 120B-parameter open multimodal agent model that activates 12B parameters per token for enterprise agents, plus two edge models.
Treasuries led a global bond rally as AI shares deepened their selloff and CoreWeave's credit risk hit a record high. US producer prices rose at the fastest pace in more than four years, adding to monetary-policy questions.
Vinci4D says its physics-driven AI software can run hardware simulations up to 1,000 times faster than conventional tools, allowing engineers to test design changes in a day instead of two weeks.
TSMC reported third-quarter revenue of $46.36 billion and a 68.7% rise in net income, topping estimates, while its shares fell 2% as an AI-related selloff hit chipmakers including Samsung and SK hynix.
October Monthly Meeting names Goldman Sachs and Starbucks as top buys, marks Broadcom for sale, reviews AI-trade holdings.
Volvo will expand its Nvidia partnership to launch a Level 2++ hands-free, point-to-point driving system by the end of the decade, using Hyperion hardware and Alpamayo software. Volvo did not disclose terms or the first model.
Data center backlash over electricity, water and local impacts is becoming a midterm issue, with Cato's Jennifer Huddleston urging the tech industry to better explain data centers' role beyond generative AI.
Meta launched Muse Spark Open, an 816B open-weight model optimized for Mac Studio with Apple's MLX.
CoreWeave credit default swaps surged to distress levels on Oct. 15, 2026, as investors soured on AI debt and AI shares fell. Nvidia CEO Jensen Huang and CoreWeave's CEO downplayed concerns, while inflation fears and Fed rate uncertainty added pressure.
Twelve state attorneys general asked FERC to authorize a 17.8-mile natural gas pipeline in New Mexico to power an Oracle data center campus for OpenAI, saying a broader environmental review could affect similar projects.
Nvidia's RTX Spark laptops launch Friday, promising full Windows apps and games on Arm after earlier Arm PC failed.
Cerebras CSO Andy Hock says agentic workflows make inference speed a requirement, and the fastest tokens will decide which AI platforms win the next phase. He cites the company’s Wafer-Scale Engine and a shift toward value per token.
TSMC reported a sixth straight record quarter and forecast 58% fourth-quarter revenue growth, as AI chip customers compete for limited wafer capacity and the foundry races to add fabs in Taiwan, Japan and Arizona.
Nvidia on Oct. 15 introduced the open-weight Nemotron 3.5 Super VL, a 120-billion-parameter model for always-on agents, plus two smaller Bolt edge models. Nvidia says the family targets faster, more customizable enterprise and local AI deployments.
Nvidia CEO Jensen Huang called on Europe to dream bigger about artificial intelligence, with Nvidia’s help, at an event in central London on Wednesday evening. He cheered “Go, go UK!” beneath stained glass windows at a church-turned upmarket restaurant. The report also referenced UK Prime Minister Keir Starmer and Peter Kyle.
The Verge reported that Microsoft and Nvidia have prepared Windows for the first RTX Spark laptops, which use Arm CPUs, Blackwell GPUs, and unified memory. Microsoft has focused on local AI agents, app emulation, gaming, and memory management.
AI-linked shares came under renewed pressure Thursday as South Korea's Kospi fell more than 7%, SK Hynix and CoreWeave weakened, and hot U.S. producer inflation added to market nerves despite record TSMC profit and a Bank of New York earnings beat.