Skip to the letters
AI Industry & Funding
-
A
SpaceX's Kessel Rack Swaps NVLink Cables for an 84-Layer Midplane
SpaceX reportedly plans a custom Vera Rubin NVL72 variant, Kessel, that replaces NVLink cables with an 84-layer PCB midplane for ground and orbital us...
-
A
SoftBank Cancels $3.7 Billion in Gas Turbine Reservations as AI Power Bets Shift
SoftBank is canceling turbine reservations that cost it $3.7 billion, shifting its AI power strategy while disputing claims it defaulted on its orders...
-
A
ASRock Ships VBIOS 1.0.6 for Radeon RX 7900 Cards With No Changelog
ASRock's new VBIOS 1.0.6 for Radeon RX 7000 cards arrives with no changelog, after 1.0.5 reportedly left some RX 7900 XT and XTX cards unbootable.
-
A
Montech King 95 3D Pro Review: Three Glass Panels, Strong CPU Cooling, Loud Fans
Montech's King 95 3D adds a top glass panel to the curved-glass King 95. We test the $199.90 Pro with seven fans for thermals, noise and build fit.
-
A
Kora's Linux Desktop Splits AI Across CPU, GPU and NPU, but Only Snapdragon X2 Elite Works Today
Kora, a Linux desktop from Cyanogen co-founder Kirt McMaster, splits local AI across CPU, GPU and NPU. Only Qualcomm's Snapdragon X2 Elite is supporte...
-
A
AMD's MEXT Aims to Halve Server DRAM by Predicting Which Memory Pages You'll Need
AMD's MEXT moves cold memory pages to flash and uses a predictive AI model to bring them back, claiming 95% of full-server performance at half the DRA...
-
A
SpaceX Plans $24 Billion Gas-Fired AI Compute Site in Starbase, Texas
SpaceX's proposed $24 billion gas-fired compute campus in Starbase, Texas, could test how much local control a company town really offers.
-
A
Fiber Optic Shortage Squeezes Ukrainian Drone Output as AI Data Centers Absorb Supply
AI data center demand has pushed fiber optic cable prices up six to 10 times, and Ukraine's drone makers are feeling it first.
-
A
Newegg's 9800X3D Bundle Undercuts Standalone 32GB DDR5 Pricing
A Newegg combo pairs the Ryzen 7 9800X3D with 32GB of DDR5-6000 and an X870E board for $1,059.98, putting the memory well below standalone prices.
-
A
Water-Cooled PC Sends Its Heat Outdoors Through a Valve-Controlled Radiator
A Reddit builder routed his water-cooled PC loop to an outdoor radiator, using an Arduino and dew-point logic to keep the room from overheating in win...
-
A
LinkedIn's Frequent Viewer Test Flags Repeat Profile Visitors
LinkedIn is testing a Frequent Viewer tag for Premium users. It shows who keeps returning to your profile, which can be useful or unsettling.
-
A
AMD's Epyc 6 Pushes Core Density and Clock Speed for AI Agent Workloads
AMD's Venice-generation Epyc 6 lineup pairs 256-core density with 5.3 GHz parts, aimed at running large numbers of AI agents on CPUs.
-
A
AMD Says Zen 6 Wasn't Built for Agentic AI, and Its Argument Is About Range
AMD's Zen 6 Venice CPUs weren't designed for agentic AI, executives say. Their pitch is that agent workloads need a range of CPUs, not one SKU.
-
A
AMD's EPYC Embedded 9006 Brings Zen 6 Cores to Chip Test and Network Gear
AMD's EPYC Embedded 9006 SP7 chips bring up to 256 Zen 6 cores to chip test rigs, firewalls and storage, with PCIe Gen 6 and fast DDR5 memory.
-
A
AMD EPYC Venice Brings Zen 6 With 27% Average IPC Gain Over Zen 5
AMD's first Zen 6 chips, EPYC Venice, show a 27% average IPC uplift over Zen 5 in data center workloads, built on TSMC's N2P node.
-
A
Alice Huang's Alleged Role in Routing Nvidia AI Chips Into China
An investigation ties Alice Huang to Megaspeed and Shanghai Century Litong, with claims of 240,000 banned chips and $7.5 billion in GPU imports.
-
A
Staera Bets on Light Inside the Chip With $4.6M Pre-Seed Round
Staera's $4.6M pre-seed round funds Lightwire, an on-chip optical interconnect it says could replace copper links between AI accelerator chiplets.
-
A
PuppyGraph Raises $10M to Let AI Agents Query Warehouse Data as a Graph
PuppyGraph raised $10M to scale its query engine, which lets AI agents traverse warehouse and lakehouse tables as one graph without migrating data.
-
A
M5 Ultra Mac Studio vs. DGX Spark: Where Local AI Performance Holds Up
A 256GB M5 Ultra Mac Studio beats a DGX Spark on agentic tasks and energy use, but for the largest models the software stack decides the winner.
-
A
Terafab's 1 Million Wafer Target Leaves Its Memory Plans Open
Terafab's reported post-2032 target of 1 million wafers a month would rival Samsung's DRAM output, but its memory process and partners remain undisclo...
-
A
Steam Shows Average Framerates From Real Players Before You Buy
Steam is adding opt-in framerate data to game pages, starting with Steam Deck. Shoppers can see average FPS from matching hardware before they buy.