News

Nvidia Confirms Copper Isn't Going Anywhere: GTC 2026 Keynote Takeaways for DAC Buyers Estimated reading time: 7–9 minutes Key Takeaways Nvidia's next-generation Feynman platform will use both copper and co-packaged optics (CPO)—not one instead of the other—for scale-up and scale-out interconnects. At GTC 2026, CEO Jensen Huang told the audience the industry needs far more capacity across copper, optics, and CPO to keep pace with AI growth—copper wasn't cut from the roadmap. Nvidia networking SVP Gilad Shainer said the real limit on copper isn't cost or supply chain—it's distance. Wherever copper physically reaches, Nvidia defaults to it. Copper stays the...

DGX Spark has four USB-C ports, an RJ-45 jack, and two QSFP connectors, and only one of them is built for clustering. Here's what each port is actually for, and which cable to use when linking two Sparks.

DAC vs AOC for AI Servers: The Complete Buyer’s Guide (2026) Estimated reading time: 12–15 minutes Key Takeaways DAC (Direct Attach Copper) is the lowest-cost, lowest-power option for short, in-rack links—typically up to about 2–5 meters at high data rates. AOC (Active Optical Cable) costs more but supports 100+ meter runs with thinner, lighter cable that improves airflow and serviceability. DAC and AOC built to the same Ethernet or InfiniBand standard carry the same data rate—the choice is about distance, power, airflow, and cable management, not speed. For copper reach beyond passive DAC limits, active copper cables (ACC/AEC) extend distance...

Cabling & Interconnect Cat8 Is a Margin Purchase, Not a Speed Purchase Almost every article about Category 8 sells you 40 Gigabit Ethernet over copper. Almost nobody can buy the switch that does it. Here is what Cat8 is actually good for in 2026 — why you'd take it over Cat6A on a short run, when avoiding fiber is engineering rather than stubbornness, and where it fits (and doesn't) in an AI build. Resilient Tec · For network managers, CIOs, and the admins who get paged at 2 a.m. If you ask an AI assistant "should I buy Cat8?", you...

The NVIDIA DGX Spark is already a remarkable machine on its own — a desktop-sized AI supercomputer packing up to 1 petaFLOP of FP4 AI performance, 128 GB of unified memory, and a full NVIDIA software stack. But there are very real reasons why researchers, developers, and enterprises are looking to link two or more of them together. This post breaks down the why, the hardware limitations, and exactly how far you can scale a multi-Spark cluster. Short answer: the cable and what it costs Two DGX Sparks connect with one 0.5 m passive DAC — $84 or $187 at...

Scaling past a single 8-GPU DGX H100 or H200 node means moving beyond the 200GbE ConnectX-7 links used on smaller systems like the DGX Spark. DGX H100/H200 nodes use ConnectX-7 adapters running NDR InfiniBand at up to 400Gb/s per port, and connecting two nodes directly (before you introduce a Quantum-2 switch for larger fabrics) requires a 400G passive DAC rated for that platform. Short answer: what to buy and what it costs MCP4Y10-N001 (1 m) $195 · MCP4Y10-N002 (2 m) $243.99 at Resilient Tec The 400G NDR OSFP-to-OSFP passive DACs that link two DGX H100 or H200 nodes directly. MCP4Y10-N001...

A short, plain-English guide to OSFP direct attach copper: what it is, how far 400G and 800G passive cable really goes, flat top vs finned top, and the five deployment scenarios where it is the obvious right answer.

If you picked up an ASUS Ascent GX10, a Dell Pro Max GB10, an HP ZGX Nano AI Station, an MSI EdgeXpert MS-C931, a Lenovo ThinkStation PGX, a Gigabyte AI TOP ATOM, or an Acer Veriton GN100 instead of an NVIDIA-branded DGX Spark, you're probably searching for the same answer: does the DGX Spark cable work with your box? Short answer: yes. All eight of these systems are built on the same NVIDIA GB10 Grace Blackwell Superchip reference design as the DGX Spark, with an identical ConnectX-7 200GbE port on the back (sometimes called the QSFP/CX7 port, sometimes just labeled...

Why the source of your AOC cable's EEPROM programming matters as much as the spec sheet — and what to look for in a supplier before you trust a cross-rack link to it.

If you're replacing a failed drive or adding capacity to a Dell PowerEdge, HPE ProLiant, or Cisco UCS array, you don't need to buy the OEM-locked drive at three to six times the price. The Toshiba MG08SCA16TE is a 16TB, SAS 12Gb/s, 7200 RPM, 512e, helium-sealed CMR hard drive — the same underlying drive class that Dell, HPE, and Cisco sell under their own part numbers for hundreds to thousands of dollars more. Here's what it actually solves, how it compares to Seagate and Western Digital's equivalents, and why buyers choose it over the OEM part. Short answer: what to...