Cisco is pushing what Managing Director Mohanad Abuissa calls a factory setup for AI infrastructure in the UAE, ditching the old model where data centers just handled individual tasks. The company rolled out its Secure AI Factory with NVIDIA, which bundles compute power, networking gear, and security layers into these modular things called AI PODs that businesses can scale up or drop wherever they need processing to happen. Abuissa says the whole point is helping companies go from testing AI projects to actually running them at scale without everything falling apart.
The factory comparison makes sense because AI workloads eat through massive amounts of data and need every part of the stack playing nice together, unlike regular...
Airtel Zambia is dropping $107 million on what they're calling their biggest network upgrade ever, and Managing Director Hussam Baday says the whole thing is about fixing congestion and getting better signals to places that need it most. They're building 406 new cell sites across the country, with 121 already up and running and the rest set to finish by next year. The focus is on schools, hospitals, and clinics where connectivity has been spotty.
The telecom company says customers should see faster downloads, clearer calls, and fewer dropped connections once everything is live. Baday mentioned that power issues have been getting better lately, which helps the tower partners keep things stable. The project responds to complaints from...
Apple pushed back mass production of its Baltra AI chip by another year after whispers suggested the thing would hit manufacturing lines sooner. Broadcom got roped into the network tech side while TSMC handles fab work on 3nm nodes, and the delay reasons remain murky. The company already stocked its data centers with M2 Ultra gear and appears ready to slot in M4-based servers for Private Cloud Compute tasks.
The Baltra setup leans toward inference work rather than training models from scratch, with estimates pointing to clusters of roughly 64 chips linked through high-bandwidth LPDDR memory instead of massive GPU farms. Cost efficiency matters more than raw horsepower for what Apple wants to accomplish.
Meanwhile, the tech giant...
ARCTIC shipped MX-7 thermal paste that handles both regular CPU mounting and direct-die applications on graphics cards, laptops, and console chips. The formula dropped thermal resistance compared to older compounds while staying non-conductive to avoid shorting components, and the thick consistency spreads through cooler pressure instead of needing a spatula.
The company tweaked the formula for better cohesiveness to fight pump-out when temps cycle repeatedly, and each tube gets an authenticity check to verify legitimate product. The redesigned syringe makes application less messy than previous versions.
Pricing starts at 6.49 euros for 2 grams and scales up to 8.49 euros for an 8-gram tube, with an option that bundles cleaning wipes...
D-Link rolled out Nuclias Unity as a license-free cloud management platform that lets IT teams wrangle switches and access points across multiple sites without paying recurring subscription fees. The system handles everything from single-location shops to massive distributed enterprises through a three-tier structure that scales from basic setups to full role-based access controls for regulated environments.
The platform tackles the usual headaches of managing gear across different branches by dumping real-time topology maps, bulk port configuration tools, and proactive capacity monitoring into one interface. System integrators can also use the tenant-based setup for service delivery, while wireless analytics track channel congestion...
Colorful launched three tiny graphics cards that crush NVIDIA's SFF-Ready spec by staying under 18 centimeters long and sticking to exactly two slots thick. The iGame GeForce RTX 5070 Mini OC leads the pack with the RTX 5060 Ti variants behind it in 16-gig and 8-gig flavors, and all three share basically the same cooling setup with a single 95mm fan pushing air through aluminum fins that stick out past the shroud.
The RTX 5070 Mini got clocked to 2557 MHz versus the stock 2512, while both 5060 Ti cards run 2632 MHz against reference 2572. Power delivery splits between a 16-pin connector rated for 300 watts on the 5070 and standard 8-pin plugs on the 5060 Ti models. Metal backplates cover the rear, and the 16 gig version cools half its...
Thermaltake dropped the TH360 V3 Ultra ARGB Sync Snow Edition liquid cooler with a cube-shaped pump block that mounts a floating 4-inch square screen running 720 by 720 pixels. The display rotates through software without physical parts and hooks into TT RGB Plus for system monitoring while ARGB lighting frames the sides. Three 120mm fans push 85 CFM each with adjustable speeds between 500 and 2,500 RPM.
The pump spins from 800 to 2,500 RPM through 46cm tubing connected to a 360mm radiator, and the whole setup handles thermal loads up to 365 watts. Socket support covers AM5, AM4, LGA1851, LGA1700, and older Intel platforms for current desktop chips.
Thermaltake skipped pricing details but positioned the cooler just under their...
ENERMAX shipped the REVOLUTION III S power supply rated at 1000 watts with ATX 3.1 compliance and a dedicated 12V-2x6 connector that pipes 600 watts straight to modern GPUs. The thing earned Platinum certification from multiple testing outfits and comes in black or white colorways with fully modular cables that have mesh sleeving. Japanese capacitors handle the internal components, and a 120mm fan with fluid dynamic bearings keeps things quiet under 60 percent load.
The semi-fanless cooling shuts off completely when temperatures drop, or the system idles lightly, and ENERMAX backed everything with a 13-year warranty. The black version goes for 180 bucks while the white model costs 190, and both hit shelves through Amazon and Newegg...
Acer and ASUS both dropped warnings that the DRAM shortage is about to mess with laptop pricing, but retail sticker prices are holding steady for the moment because companies locked in supply contracts that shield them from spot market chaos. Memory used to run about 8 to 10 percent of a PC build cost, and the 30 to 50 percent spike only pushed total costs up by maybe 2 or 3 percent since manufacturers got protection from long-term deals.
The real pain hits when those contracts reset around mid-2026, and that is when shoppers will actually see price bumps show up on spec sheets. Budget laptops are probably keeping 8 gigs of RAM and 256 gig SSDs while squeezing vendor margins, mid-range gear might get downgraded parts with pricier...
