The 5-step pipeline.
How energy turns into a deployed, financed, offtake-matched cluster, and what SLYD does at every step.
How SLYD works →Platform
SLYD Cloud
New and recovered GPU systems.
NVIDIA and AMD systems through documented manufacturer and qualified channel supply, with financing and deployment coordinated on the same platform.
Explore GPU hardware →By accelerator
Infrastructure and services
Compute, hardware, and power in one book.
Browse available GPU capacity by accelerator, configuration, region, and price, or bring supply to qualified demand.
Open marketplace →Compute
Bring supply
For buyers
Start with an indicative structure.
Tell us deal size, structure, and offtake. Any range is preliminary and subject to underwriting, diligence, and documentation.
Open Configure →For lenders
GPU market trends and deployment playbooks.
Infrastructure best practices, hardware comparisons, and industry analysis from the SLYD team.
Read the blog →Company
Supermicro AI servers
Supermicro's distinguishing move is publishing the same accelerator platform in several physical formats, across two rack standards and both cooling methods. When the building is the constraint, that breadth is the reason to look here.
What are Supermicro AI servers?
Supermicro provides a broad portfolio of GPU servers, rack-scale AI systems, and air- and liquid-cooled platforms. The correct configuration depends on accelerator support, server density, networking, storage, power, cooling, management, workload, and the exact commercial and service terms available for the transaction.
Which rack standard does the site use
With Supermicro this question comes before the accelerator question, because it removes half the catalog either way. Supermicro publishes its liquid-cooled HGX B300 system in both standards, and they are not interchangeable.
4U liquid-cooled HGX B300
Built for standard 19-inch EIA racks, which is what most enterprise and colocation estates already have. Supermicro publishes this as a front-I/O system using its DLC-2 direct liquid cooling technology.
2-OU liquid-cooled HGX B300
Built to the 21-inch OCP Open Rack V3 specification, with blind-mate manifold connections and a modular GPU and CPU tray architecture. Requires an ORV3 estate, and delivers substantially higher density in return.
Supermicro states that the 2-OU OCP liquid-cooled HGX B300 system enables up to 144 GPUs per rack, and that it carries eight NVIDIA Blackwell Ultra GPUs at up to 1,100W TDP each. Those figures describe that configuration in an ORV3 rack. They are not a portfolio-wide capability, and reaching that density depends entirely on the facility supplying matching power and cooling.
What Supermicro publishes for AI systems
Reproduced from Supermicro's published materials. Supermicro's catalog is large and changes frequently, so this page describes the shape of the portfolio rather than reproducing a model list that would be stale on arrival.
| Category | What Supermicro publishes | What to confirm |
|---|---|---|
| NVIDIA Blackwell portfolio | Solutions spanning HGX B300, HGX B200, and GB200 NVL72 | The exact model number, GPU count, and cooling method for the system being quoted |
| Liquid-cooled HGX B300 | 4U for 19-inch EIA racks and 2-OU for 21-inch OCP ORV3 racks | Which rack standard the site uses, and whether facility water is available |
| Rack-scale AI | Generative AI SuperCluster offerings | What is included in the rack-scale scope: networking, storage, integration, and validation |
| Direct liquid cooling | DLC-2 technology, published with a stated heat capture figure for a named configuration | The heat capture figure for the exact configuration, and how the remaining heat is handled |
| Networking integration | Integrated NVIDIA ConnectX-8 SuperNICs on published HGX B300 systems | Adapter count, port speed, and whether optics and cabling are in scope |
SLYD does not publish a configuration count for any manufacturer. A number like that is only meaningful if it is generated from a current catalog record with a review date, and it goes stale faster than a page can be maintained.
Reading a heat capture figure properly
Supermicro states that its DLC-2 technology, on the front-I/O liquid-cooled 4U NVIDIA HGX B300 8-GPU system, captures up to 92 percent of the heat generated by server components.
Two things follow from reading that carefully. It is an "up to" figure for one named configuration, so it does not transfer to a different Supermicro system or to a different manufacturer's direct liquid cooling. And the remaining share of the heat is still real: it leaves the server as air and the room still has to handle it. A facility plan that budgets for zero air-side load because the system is liquid cooled will be wrong.
Supermicro also publishes a data center power savings percentage for this technology. It is not reproduced here, because the baseline and measurement conditions behind it were not retrievable at review time, and an efficiency percentage without a stated baseline is not a usable planning input. The cooling page covers how to evaluate these claims.
Supermicro products and SLYD process are separate things
Manufacturer products and services
System designs, cooling technology, rack-scale offerings, and any manufacturer service are Supermicro's, with scope and eligibility defined by Supermicro for the specific product and contract.
Sourcing process
SLYD evaluates a stated requirement, identifies which documented sourcing paths are available for the requested system, and confirms configuration, seller authority, provenance, condition, availability, and commercial terms for the specific transaction.
SLYD is an authorized reseller for Dell Technologies, HPE, Supermicro, Lenovo, Cisco and IBM, and sources through authorized distribution including Arrow, Ingram Micro and TD SYNNEX. Configuration availability and lead times are quoted at time of order. SLYD makes no claim of a privileged supply position, and no standing lead-time, savings, or support commitment for this or any manufacturer.
To request a configuration
- Workload and scale
- Preferred accelerator, if already chosen
- Node count and expected growth
- Deployment location
- Rack standard in use, 19-inch EIA or OCP ORV3
- Available rack power
- Whether facility water is available for direct liquid cooling
- Target timeline
- Compliance and export constraints
- Capital structure preference
Supermicro questions
What makes Supermicro different from the other server manufacturers?
Breadth of form factor. Supermicro tends to publish the same accelerator platform in several physical formats, including both 19-inch EIA and OCP ORV3 rack standards and both air and liquid cooling. That is useful when the facility, rather than the accelerator, is the constraint, because it increases the chance a deployable format exists for the site.
Which Supermicro systems support NVIDIA Blackwell accelerators?
Supermicro publishes an NVIDIA Blackwell portfolio spanning HGX B300, HGX B200, and GB200 NVL72 solutions. For HGX B300 specifically, Supermicro publishes liquid-cooled systems in a 4U form factor for 19-inch EIA racks and a 2-OU form factor for 21-inch OCP ORV3 racks.
How dense can a Supermicro GPU rack get?
Supermicro states that its 2-OU OCP liquid-cooled NVIDIA HGX B300 system enables up to 144 GPUs per rack. That figure belongs to that specific configuration in an OCP ORV3 rack and should not be applied to the portfolio generally. Achieving it also depends entirely on the facility supplying the matching power and cooling capacity.
What is Supermicro DLC-2 and what does it capture?
DLC-2 is Supermicro's direct liquid cooling technology. Supermicro states that on the front-I/O liquid-cooled 4U NVIDIA HGX B300 8-GPU system it captures up to 92 percent of the heat generated by server components. That is a Supermicro figure for that named configuration, and the remaining heat still has to be handled by facility air.
What is the difference between the 19-inch and OCP versions?
The rack standard, which determines whether the system can be installed at all. The 4U liquid-cooled HGX B300 system is built for standard 19-inch EIA racks, while the 2-OU version is built to the 21-inch OCP Open Rack V3 specification. An estate standardized on 19-inch racks cannot take ORV3 equipment without a rack change.
Does a Supermicro product named here mean SLYD has it in stock?
No. Supermicro system names and figures are reproduced from Supermicro's published materials to explain the architecture choices available. Whether a specific configuration can be obtained, in what quantity, for what geography, and on what terms is confirmed at the transaction level.
How are lead time, warranty, and commercial terms verified?
At the transaction level, for the specific equipment, quantity, and geography, with the confirmation date and scope stated. SLYD publishes no standing lead-time estimate for any manufacturer, because lead times move with product, configuration, supply position, and destination.
What information is needed to request a Supermicro configuration?
The workload and scale, preferred accelerator if already chosen, node count, deployment location, the rack standard in use, available rack power, whether the site supports facility water for direct liquid cooling, target timeline, and any compliance or capital constraints.
Sources for the Supermicro information on this page
- Supermicro, announcement of 4U and 2-OU OCP liquid-cooled NVIDIA HGX B300 solutions. Reviewed 18 August 2026. supermicro.com press release
- Supermicro, NVIDIA accelerator solutions. Reviewed 18 August 2026. supermicro.com NVIDIA solutions
- Supermicro, GPU systems. supermicro.com GPU systems
- Supermicro, liquid cooling solutions. supermicro.com liquid cooling
- Supermicro, Generative AI SuperCluster. supermicro.com AI SuperCluster
System descriptions, density figures, and cooling figures are Supermicro's, reproduced with the configuration they apply to. They are not SLYD guarantees of configuration, performance, or availability. Supermicro is a trademark of its owner and is used here descriptively.
Request a Supermicro configuration
Start with the rack standard, available power, and whether facility water exists. Those three answers narrow a very large catalog quickly.
Page updated: August 18, 2026