Aug 25, 2026InfrastructureLaunchInfrastructureOracle Adds AMD GPU Operator to OKE for Broader GPU Lifecycle ManagementThe optional enhanced-cluster add-on broadens Oracle’s earlier device-plugin support into driver, health and configuration management, while leaving teams responsible for supported stack combinations and component sizing.3 min
Aug 25, 2026InfrastructureGovernment actionInfrastructureTrump Backs Data Centers as Paxton Proposes Texas Limits on Grid Costs and Rural SitingThe dispute is shifting from whether AI infrastructure brings investment to who pays for the power it needs, and where the facilities can be built.3 min
Aug 25, 2026InfrastructureBenchmarkInfrastructureOpenAI’s Jalapeño Claims 1.5–1.9x More AI Work Per Watt, Faces 2027 Scale TestThe custom chip gives OpenAI an early efficiency claim against available Nvidia systems, but production qualification, broader workloads and a small initial rollout will determine whether it becomes a meaningful serving platform.4 min
Aug 25, 2026InfrastructurePartnershipInfrastructureCisco Adds Supermicro Rack-Scale Systems to Its NVIDIA AI Factory for October RolloutThe partnership turns Cisco’s AI infrastructure offer into a more complete rack-to-fabric package, combining dense GPU servers, networking and liquid cooling for customers building larger AI clusters.2 min
Aug 25, 2026InfrastructureEnterprise adoptionInfrastructureKimi.ai Uses TiDB for One-Second Agent Databases and Persistent Development StatePingCAP’s deployment pitch is built around instant database creation and durable agent context, but its headline capacity, cost and speed figures have not been paired with measured Kimi.ai operating results.3 min
Aug 25, 2026InfrastructureGovernment actionInfrastructureTwo Planned English Data Centres Seek Gas Power, With a 4.577m-Tonne Emissions EstimateThe dispute is not over whether the projects need vast power, but whether their eventual operating levels and route to electricity can fit the UK’s climate plans.3 min
Aug 25, 2026InfrastructureResearchInfrastructureOracle Puts Access Filters Before Agent Search—and Reranking Adds 2.2 SecondsThe proposed database-centered pipeline separates a correctness requirement from a relevance choice: access scope must constrain candidates first, while vector, lexical, fusion, and reranking stages still have to prove their value on each workload.3 min
Aug 24, 2026InfrastructureInfrastructureGartner Lifts 2026 Chip Forecast to $1.6T as AI Memory Demand Tightens SupplyThe forecast shifts attention from accelerators alone to the memory and supply-chain capacity needed to run bigger AI systems—and leaves technology buyers facing elevated prices into 2027.3 min
Aug 24, 2026InfrastructureEarningsInfrastructureNvidia Falls More Than 2.2% Before Earnings as Seven-Day Slide ExtendsThe coming results must answer a market already weighing semiconductor weakness, rising memory costs and questions around the AI infrastructure boom.2 min
Aug 24, 2026InfrastructureFundingInfrastructureXPeng Raises $900M+ at $6.3B for IRON, With 2027 Sales Still AheadThe financing funds chips, models, data and factories in one push. Its commercial proof still depends on production and customer deliveries that XPeng has yet to begin.3 min
Aug 24, 2026InfrastructureLaunchInfrastructureNVIDIA Says MaxLPS Can Add Up to 40% More Rubin GPUs Within a Fixed Power BudgetThe suite treats unused rack headroom, workload tuning and cooling power as capacity to recover. Its biggest capacity claim is still a projection, while its control software remains in preview.2 min
Aug 24, 2026InfrastructureLaunchInfrastructureAWS Brings Ray Into SageMaker HyperPod With Recovery Tools and Tiered Cache on EKSThe launch moves several operational tasks behind SageMaker Studio while tying Ray teams to a HyperPod-on-EKS setup and its required add-ons.3 min
Aug 24, 2026InfrastructureOpen releaseInfrastructureMeta Sends MetaRoCE to OCP for Loss-Tolerant AI EthernetThe proposed RDMA transport treats out-of-order arrival and packet loss as normal network conditions, shifting more control to endpoint NICs. Meta’s reported results are promising, but hardware interoperability remains the consequential test.3 min
Aug 24, 2026InfrastructureFundingInfrastructureAlibaba Seeks $10.2B for AI Infrastructure as Hong Kong Shares Fall 10%The discounted equity sale funds a larger AI buildout, but it also raises the return threshold for a company already spending more heavily while profit has fallen.4 min
Aug 24, 2026InfrastructurePartnershipInfrastructureNVIDIA Invests in Lancium, Linking DSX Designs to a 15+ GW AI Campus PipelineThe deal joins capital, AI-system designs and claimed power-secured sites. Its commercial case now depends on turning a development portfolio into operating GPU campuses with the promised power efficiency.3 min
Aug 24, 2026InfrastructureLaunchInfrastructureNVIDIA Adds BlueField-4 Scale-In for 800 Gb/s AI-Factory Infrastructure OffloadThe new layer targets a growing bottleneck around GPU clusters: moving, securing and governing data without consuming the host resources assigned to AI workloads. Its performance case remains NVIDIA-supplied.3 min
Aug 24, 2026InfrastructureOpen releaseInfrastructureAWS Backs ARD Open Spec to Federate AI-Agent Discovery Across EnvironmentsThe Apache 2.0 specification is designed to make agents, MCP servers, tools and APIs discoverable across separate catalogs without moving their governance into one central system.3 min
Aug 24, 2026InfrastructureResearchInfrastructureUC San Diego Will Test $8.48M System to Make 2 MW of AI Compute Grid-ResponsiveThe planned project couples a direct-to-DC power path with software that can defer some computing work. Its practical value rests on whether both systems perform together under live research workloads.3 min