Top stories in Chips & Infrastructure
EUV lithography supply is a hard bottleneck on global AI compute expansion.
ASML raised its full-year sales outlook and forecast Q4 revenue above expectations as AI-driven chip demand surges, but warned it will be unable to meet customer demand in 2027 and 2028.
ASML
More:
Reuters ,
Euronext Markets: Real-time Stock Market Data | live ,
TradingView ,
Central Oregon Daily ,
KSL News ,
Investing.com ,
U.S. News - Money ,
Bellingham Herald ,
Bradenton Herald ,
WTVB ,
TradingKey ,
Hacker News ,
Yahoo Finance ,
Barron's ·
💬 129 on HN
Brings petaflop local AI compute to consumer laptops, enabling on-device models and creation.
Nvidia's RTX Spark superchip combines gaming PC, creative studio and personal AI supercomputer roles, with up to 128GB unified memory and 1 petaflop of AI compute. DLSS 5 updates ship alongside.
Nvidia On sale
More:
The Verge ,
The Hindu ,
Blockchain News ,
Gadget Pilipinas ,
TechPowerUp ,
Inven Global ,
DLCompare.com ,
GameGPU ,
igor´sLAB ,
The FPS Review ,
Club386 ,
PCMag ,
Yahoo Tech ,
MMORPG.com ·
💬 1 on HN
Inference competition between AMD and Nvidia affects AI compute pricing and supply.
Analysis outlets examine whether AMD is a real threat to Nvidia or just a pricing lever, after SemiAnalysis found AMD challenging Nvidia in AI inference workloads.
AMD
Shows Huawei converting export controls into durable share gains inside China's AI chip market.
Nvidia's share of China's AI chip market has fallen to about 34% as Huawei closes the gap, Reuters reports, underscoring how export restrictions and local alternatives are reshaping the market.
Nvidia
More:
TradingView ,
The Lufkin Daily News ,
Reuters ,
1470 & 100.3 WMBD ,
BigGo Finance ,
Investing.com ,
Mexico Business News ,
36Kr ,
bloomingbit ,
digitimes ,
TechNode ,
GuruFocus
Reports warn that the AI memory-chip boom now faces a new risk: too much competition among its most valuable players, which could pressure pricing and margins.
Anthropic's first Japan compute site signals AI infrastructure shifting to regional build-outs.
Anthropic is in final talks with NTT Data Group to build its first data center in Japan, Nikkei reported; the facility could draw up to 500 megawatts, making it among Japan's largest.
Anthropic
AI infrastructure firms' shares fell as markets priced in higher odds of US interest rate hikes, Press reports.
DA Davidson's Gil Luria told Bloomberg that Nvidia's ecosystem and chip demand could support growth through 2030, but warned debt-heavy data-center builders may face serious pressure.
War targeting major data centers raises the physical security stakes for AI compute.
Reuters reports Ukrainian drone strikes hit Russia's biggest data center, an attack on infrastructure that services both military and civilian computing needs.
Rising rack power densities shape data center design and cooling requirements for AI.
Reports say new Nvidia AI racks support 250 kilowatts per rack, while IBM and Siemens highlight a 7-angstrom AI chip research collaboration. Nvidia shares fell 2.9%.
Nvidia
More in Chips & Infrastructure
Axionic, an AI chip startup founded by a former Qualcomm executive, raised an $80M pre-seed round, per Dealroom.
Axionic
CoreWeave co-founder says the Nvidia-backed cloud provider is courting insurers' vast capital for loans as it funds data center build-out.
CoreWeave
Estonia's Skeleton Technologies became a unicorn after raising investment from Siemens to manufacture energy storage systems for AI data centers in the US.
Skeleton Technologies
SK hynix detailed H-TPU, tiny in-house CPU cores that offload routine I/O from Arm CPUs in its next enterprise SSD controllers as drive speeds rise.
SK hynix
At OCP Global Summit 2026, Compal unveiled next-generation Nvidia-based AI platforms and Liteon showcased Nvidia's 800 VDC power architecture and thermal management portfolio.
Nvidia
Mistral AI models are running aboard the first satellite of a planned 50-satellite AI constellation, putting models in orbit.
Mistral AI Closed
Nikkei Asia reports SB Energy, SoftBank's US renewable energy arm, is doing the groundwork for Masayoshi Son's AI push by building solar and battery capacity to power data centers.
SB Energy
Analysis piece on how Singapore has become a key node in the global AI supply chain, covering chip, data center and logistics links between the US, China and Southeast Asia.
Cisco and Nvidia are teaming on a new optical pluggable called USFP that targets 12.8 Tbps per module, far above today's 1.6 Tbps optics, for AI datacenter networking.
Cisco
Ark's Cathie Wood says she avoids traditional memory-chip makers due to the volatile cycle, favoring Nvidia and AMD as the AI revolution is 'just beginning.'
Ark Investment Management
Nvidia is prepaying for US chip-testing capacity as KYEC expands by $1.4B, with the two eying a Texas site for a US AI chip testing plant.
Nvidia
Investing.com examines why Nvidia shares fell, tying the move to broader AI trade and rate concerns.
Nvidia
Zesty released autonomous GPU resource management that fractions CUDA-native GPUs, aiming to run more workloads per GPU while protecting performance.
Zesty
Asian data center operator Princeton Digital Group is seeking a syndicated loan of up to $1.2 billion to fund its Malaysia buildout, as AI infrastructure borrowing grows.
Princeton Digital Group
Hitachi Energy CEO Andreas Schierenbeck discusses whether electricity infrastructure can expand fast enough to support the AI boom and industrial electrification.
Hitachi Energy
China's integrated circuit exports rose 136% year on year in September to US$45 billion, outpacing 108% growth for the first nine months, on AI-driven demand and higher chip prices. September volume reached 32.7 billion units.
Intel is pitching an all-in-one chip aimed at closing the commercialization gap for humanoid robots, positioning it as a 'robot brain' for physical AI.
Intel
Reports say Nvidia halted RTX 5090 production, while Japanese retail prices for RTX 50-series cards jump up to 144%, with the RTX 5080 passing $2,400.
Nvidia
Key facts from TradingView: Nvidia has shipped 678,000 AI accelerators to China and holds a Doctolib stake, with pre-market shares dipping.
Nvidia
Vast Data argues stitched-together AI pipelines struggle to keep sensitive embeddings, permissions and source data in sync, positioning unified architecture for sovereign AI inference.
Vast Data
ScaleFlux argues smarter SSD controllers can address the AI memory bottleneck as GPUs sit idle waiting for data and NAND flash takes on more processing work.
ScaleFlux
Semiconductor shares fell, dragging Korea's Kospi lower as Samsung and SK Hynix declined, after Broadcom's chief warned of a 'casino' mentality and investors fretted over the scale of AI investment.
Digitimes reports AI accelerator growth is shifting from volume toward value as custom ASICs move upmarket, changing the competitive landscape for chip vendors.
Jim Cramer said US chip stocks are 'joined at the hip' with SK hynix, arguing investors should look beyond Nvidia when weighing the AI memory supply chain.
Reuters reports US chip stocks extended their declines, hit by a China memory report and jitters over AI spending.
CyberSecurityNews reports a flaw in Nvidia's DCGM Exporter that exposes GPU servers to resource exhaustion attacks, potentially disrupting AI infrastructure monitoring.
Nvidia
Cisco and Nvidia developed a new pluggable optical module for AI data centers, targeting high-speed interconnects with four-times USFP density so systems wait less on data.
Nvidia
Chinese tech companies purchased more than a million Nvidia gaming chips in a single year, repurposing hardware never designed for advanced AI to meet soaring compute demand.
Nvidia
CYANIS AI launched Sovereign Inference Cloud, a service for running AI inference inside a customer's own jurisdiction and infrastructure. Pricing and availability were not disclosed.
CYANIS AI
A roundup reports Microsoft is accelerating its in-house AI chip challenge while Google and Apple deepen their own AI hardware and product efforts.
Microsoft
Intel's D-series data center GPU, codenamed Crescent Island, is named D270 and supports up to 480 GB of LPDDR5X memory, positioning it as an Nvidia rival for AI workloads.
Intel
A report details Nvidia's PTFE strategy, describing the fluoropolymer material approach as a hidden enabler behind its next-generation AI server racks and high-speed interconnects.
Nvidia
Satlyt, which aims to be the Android of orbital computing with open software that works on many companies' satellites, launched its software on a SpaceX rocket alongside Google's hardware after both SpaceX and Google declined to back it.
Satlyt Open
Microsoft plans a multi-gigawatt deployment of next-gen Azure Maia AI accelerator racks, with 96-chip configurations promising seven times its predecessor's token throughput and production deployments due in 2027.
Microsoft
The Information argues Applied Materials and Lam Research are overlooked chip equipment opportunities as AI chip designer, manufacturer and equipment stocks have soared.
Nvidia-backed data center firm Firmus scrapped its Australian IPO citing market conditions and is now seeking up to $3 billion, per reports.
Firmus
Builders ranked Bitdeer AI cheapest for DeepSeek V4 Flash API access on Artificial Analysis. Bitdeer also secured 67 MW in Malaysia for Nvidia AI.
Bitdeer Paid
XMEMS began mass production of chips that cool AI glasses, smartphones and data centers, targeting thermal limits in compact AI hardware.
XMEMS
Samsung, SK Hynix and CXMT shares fell sharply as investors doubt whether the AI-driven memory profit boom can last, even as soaring prices lift quarterly earnings.
Chinese AI chipmaker Iluvatar CoreX debuted the Tiangai 300 IXL144, a supernode system that fits 144 GPUs into a single cabinet, unveiled at the 2026 China International event.
Iluvatar CoreX
Navitas shares rose about 5% after the power-chip maker announced an AI infrastructure partnership with Nvidia.
Navitas
AWS engineers detail fused Mixture-of-Experts kernel work on SageMaker AI: Swap-AB for WGMMA utilization, Triton backend routing for FP8 quantization and cutting JIT recompilation, for up to 10% higher throughput.
AWS Closed
An industry report explains why semiconductor fabrication plants require robots that make almost no mistakes, given the precision and cleanliness demands of chip manufacturing.
Nexperia has expanded its gallium nitride portfolio aimed at robot drives, targeting smaller, more efficient power electronics for motors used in robotics.
Nexperia
Alice & Bob has decoded LDPC-cat codes on GPUs using NVIDIA's NVQLink, a step toward practical quantum error correction with AI hardware.
Alice & Bob
French cloud provider Scaleway plans to open an Nvidia-powered AI computing site in Copenhagen, expanding European GPU capacity.
Scaleway
TrendForce reports Microsoft's Maia 300 will fuel HBM4 demand while Intel bets on LPDDR5X memory, as AI chipmakers diverge on memory strategies.
Microsoft
Anthropic CEO Dario Amodei warned of a 'huge shortage' of AI chips and has reached out to Elon Musk about compute capacity, Firstpost reported.
Anthropic
Nvidia will no longer ship new Windows 10 drivers for RTX GPUs, per reports, pushing users toward Windows 11 for continued driver support.
Nvidia
An Economic Times video breaks down how China is structuring its AI chip strategy, covering the roles of DeepSeek, Huawei and Nvidia in the domestic compute push.
Chips & Infrastructure archive →