MSI Packed Four EPYC CPUs Into a Single Rack Unit — Because Air Cooling Just Said No

At Computex, while the big processor vendors are still keeping their 2026 roadmaps under wraps, the OEMs are getting increasingly desperate to show off the hardware they've built around them. MSI's booth included something that quietly caught my attention: a 1RU liquid-cooled chassis that holds two dual-socket server nodes, each running an upcoming AMD EPYC "Venice" processor. That means four complete Venice CPUs and 64 RDIMMs crammed into a single rack unit. The whole thing is called the CD182-S6091-X2 (DLC — "Direct Liquid Cooled"), and it's the kind of machine that exists because air cooling has officially given up.

What makes this worth paying attention to isn't just the density. It's the platform shift. Venice brings a new 16-channel DDR5 memory bus per socket, PCIe Gen6 lanes, and TDPs that AMD hasn't yet disclosed but are clearly pushing past the 500-watt mark that fifth-generation chips already hit. MSI solved the thermal problem by putting both CPUs on their own cooling loops, with the memory on a separate loop — 32 RDIMMs per node, all liquid-cooled, all on a single 1OU2N board. There's also an ASPEED AST2700 BMC (notably the newer generation), an OCP 3.0 NIC slot, FHHL and HHHL PCIe slots, and four E1.S SSD bays. The whole chassis runs on a 48V DC busbar — no power supply, just raw plumbing and forced-air for the bits that don't need it. The rack version, the ORv3, holds 28 of these nodes in 44U and can pull 100kW through a Chicony power shelf and an Auras CDU.

Source article image
Source image 1

The practical story here is straightforward: as server CPUs push past the air-cooling ceiling, the entire rack ecosystem has to change. It's not just the cold plates and loops — it's the power architecture (busbar instead of individual PSUs), the storage form factors (E1.S over U.2), the BMC generation, the OCP NIC slots replacing standard PCIe. Think about the operational cost: a single MSI ORv3 rack holds 28 compute nodes and 112 Venice CPUs, consuming 100kW through a 55kW Chicony power shelf and an Auras CDU that connects to a facility's liquid-to-liquid cooling loop. That's not a server you plug into a standard outlet. It's a facility-grade deployment that needs chilled water lines, busbar power distribution, and

Source article image
Source image 2
a floor plan that accounts for the plumbing. By the time Venice ships later this year, anyone planning a new deployment is going to need to decide whether their data center is actually ready for liquid-cooled rack infrastructure, or whether they're going to be stuck on older platforms that still play nice with fans and regular power strips.

Sources

Comments

Post a Comment

Popular posts from this blog

AI Is Starting to Feel Less Like a Gadget and More Like Infrastructure

Image
Modern AI doesn't replace infrastructure — it runs on top of it. (Photo: Unsplash) For a while, AI coverage had the energy of a mall demo kiosk: very shiny, slightly chaotic, and usually one prompt away from embarrassment. Lately, though, the more interesting story is less about flashy demos and more about infrastructure. A useful CNBC piece argued that AI is not simply going to vaporize enterprise software overnight, and that sounds about right. In real companies, software does not disappear just because a model can summarize a meeting or write a decent SQL query. What actually happens is messier and more practical: AI starts sneaking into the plumbing. It shows up in search, support workflows, procurement tools, analytics dashboards, and data pipelines. The real winners may not be the loudest chatbot wrappers, but the vendors that make enterprise systems easier to operate, easier to query, and slightly less soul-crushing for the humans trapped inside them. That is less cinematic ...
Read more

When Two AI Bots Finally Learned to Talk in Discord

I spent part of the day doing something that sounds like the setup for a bad joke: getting two local AI assistants to talk to each other in the same Discord channel. Not through a web UI. Not by bouncing prompts manually between windows like a human message queue. In the actual shared channel, where both could see the conversation and react to each other. The funny part is that the problem was not intelligence. It was manners. Both bots were defensive by default around bot-authored messages, which is the sensible setting if you do not want your infrastructure turning into a recursive support group. The downside is obvious: if both assistants treat other bots as suspicious background noise, they will never coordinate on anything more useful than silence. The fix was to make both sides accept bot-authored messages only when they were explicitly mentioned. That detail matters more than it sounds. Blanket bot acceptance is how you end up with two enthusiastic systems discovering each oth...
Read more

A CISA Contractor's GitHub Repo Held 844 MB of Secrets — and No One Closed the Door

Image
There is something almost poetic about the US government's premier cybersecurity agency — the one whose job is literally to plug holes in critical infrastructure — getting outsmarted by a contractor who treated a public GitHub repository like a digital junk drawer. The "Private-CISA" repo, created November 13, 2025, sat publicly for six months containing 844 megabytes of plaintext passwords, AWS GovCloud administrative credentials, Kubernetes manifests, ArgoCD application files, Terraform infrastructure code, CI/CD build logs, and internal deployment documentation. Guillaume Valadon at GitGuardian flagged it on May 14 after his automated scanning picked up the exposure. The commit history told the whole story in plain sight: the account owner had explicitly disabled GitHub's default secret-scanning protections, pushed plaintext credentials stored in CSV files, committed full backup archives into git history, and used easily guessed passwords like "platform-name-2...
Read more