Infrastructure and hardware comprise the physical compute systems, specialized chips, and data center components that power artificial intelligence workloads. Truncated covers chip architectures, supply chains, energy requirements, and the hardware deployments that enable large scale computing.
AI-generated summary AMD networking executive Soni Jiandani says that artificial intelligence infrastructure is too complex for a single company to own. Jiandani states that Ethernet remains a constant foundation that supports multiprotocol and multipathing for customers such as Microsoft and Oracle.
AI-generated summary Chinese artificial intelligence developers expect to receive only a small fraction of the advanced chips ordered from NVIDIA Corp over the coming months. Alibaba, Tencent, Bytedance, and other Chinese AI and cloud majors ordered about six million of Nvidia chips worth seventy billion dollars according to Bloomberg.
AI-generated summary Equinix reported record third-quarter bookings that reinforced Mizuho Securities' bullish outlook on the data centre operator as artificial intelligence infrastructure demand outpaces capacity. Mizuho maintained its Outperform rating and $1,275 price target while noting that annualised bookings crossed $500 million for the first time.
AI-generated summary Facing long delays for grid connections, some artificial intelligence data centers are turning to converted jet engines to generate their own electricity quickly. Companies including xAI and OpenAI have utilized the rebuilt turbine units to power sites, while other firms navigate lengthy waits for traditional power supply.
AI-generated summary FlyDSL allows developers to write GPU programs in Python and compile specialized kernels on demand for LLM serving. ATOM integrates FlyDSL mono decode for models including MiniMax-M3, GLM-5.2, Kimi-K3, and DeepSeek-V4.1.
AI-generated summary Engineering at Meta explains that F3 Lightning was developed to allow every feature to be released on its own schedule while preserving extraction efficiency. Engineering at Meta notes that the system compiles each feature independently and consolidates shared work at startup using identifiers based on the computations themselves.
AI-generated summary Corning is investing $585 million in expanding manufacturing in North Carolina to meet increasing demands for fiber-optic cables from companies such as Amazon, Meta, and Nvidia. Relativity Networks developed a hollow-core fiber called ChronoCore and partnered with Prysmian to scale production and connect data centers faster.
AI-generated summary OpenAI CEO Sam Altman stated on a podcast that keeping limited DRAM in personal computers used five percent of the time is quite terrible for global allocation. Microsoft executive Satya Nadella instead promotes unmetered intelligence with models running locally on Windows PCs.
AI-generated summary Seoul National University researchers announced TACToC and mzCache to reduce AI data traffic and response delays. The systems reduce communication costs and shorten the time users wait for a response when returning to an AI application.
AI-generated summary Yucca Growth Infrastructure released its right to a deadline for action on an air permit regarding Oracle Corp.'s data center in New Mexico. State environmental regulators receive additional time to review the application rather than facing a denial of the permit.
AI-generated summary AMD CEO Lisa Su announced that the chipmaker plans to selectively invest with lead partners to help secure data center land and power infrastructure for planned deployments into 2027. The company reported record data center segment revenue of $8.5 billion as part of $13.4 billion in Q3 revenue during a sustained artificial intelligence boom.
AI-generated summary IBM released open-source software built with Red Hat and Google to help companies cache data and reduce repetitive AI reading tasks. IBM states that caching and optimizations allow the same hardware to handle seven times more work while managing costs.
AI-generated summary Cerebras makes available its third and fourth generation systems, which are built to operate reliably and outlast their economic life by many years. Cerebras says that less than half of one percent of all CS-3 wafers deployed have been decommissioned since the introduction of the CS-3 in 2024.
AI-generated summary Neocloud Lambda announced plans to develop a data center campus consisting of four buildings in Lowndes County, Alabama. Lambda says the development will require an investment of $33.1 billion and generate 1,000 jobs for the duration of the project.
AI-generated summary PlatinumGames and Nintendo Switch exclusive Astral Chain has been released in a native PC port utilizing NVIDIA DLSS 5 neural rendering. The video shows that neural rendering enhances the environments with more detailed textures and slightly moodier lighting for the game.
AI-generated summary AMD announced that FSR Upscaling 4.1.2 will bring machine learning upscaling to Xbox Series X and Series S consoles for developers. AMD explains that developers can use the new console implementation to reduce upscaling overhead on hardware with limited GPU resources.
AI-generated summary Renewable Innovations announced it has joined a US DOE-backed research project to develop fuel cell power systems for AI data centers. The initiative will test how tightly coupled fuel cell and compute-rack architectures cope with the dynamic power profiles of AI workloads and includes a field demonstration at the Sky Data Centers site in Santa Rosa, California.
AI-generated summary Amazon, Microsoft, Alphabet, and Meta announced plans to spend at least $720 billion on capital expenditures this year alone, with most funds directed toward artificial intelligence infrastructure. The spending plans triggered a sharp selloff in tech stocks as investors question when they might expect to see a return on investment from artificial intelligence.
AI-generated summary Nscale released new research surveying 500 United States AI-native leaders regarding how infrastructure choices shape product velocity, engineering capacity, and compliance. Nscale says the findings show that infrastructure has moved from a background operational concern to a strategic decision that shapes scaling capabilities.
AI-generated summary Taiwanese chipmaker Macronix has rolled out a business wind-down for its ROM operations to focus on flash memory production instead. David Gibson expects the move to nudge more games toward digital downloads as the company stops developing new ROM products.
The top AI news in your email delivered each morning. Choose categories