The companies that design, manufacture, integrate or sell complete AI servers and racks: branded server makers (OEMs), original design manufacturers (ODMs) and contract manufacturers (EMS) building on platforms such as NVIDIA’s GB300 NVL72. This group is not a physical part; a builder is listed for its own servers and racks, and under a part only when separately documented as making it. Hon Hai expects its AI-rack shipments to more than double in 2026 (Aug 12, 2026).
OEMs such as Dell, HPE, Lenovo and Supermicro sell branded systems with services to enterprises and clouds. ODMs such as Foxconn Industrial Internet, Quanta, Wistron, Wiwynn and Inventec design and build to large cloud customers’ specifications, often delivering tested L10 servers and L11 racks. EMS firms such as Sanmina and Jabil manufacture to customers’ designs. Margins differ: Wiwynn reported an 8.3% gross margin for 2025, Supermicro 17.5% for its June 2026 quarter.
Do builders report how many NVL72 racks they ship?
No. Each reports differently: Dell gives AI-server orders, revenue and backlog; Hon Hai and Foxconn Industrial Internet give rack or cabinet shipment growth; Wistron gives an AI-infrastructure share of revenue. None breaks out NVL72-class racks by customer, so rack counts and market shares in the press are treated as reported and never combined.
Hon Hai and its listed subsidiary Foxconn Industrial Internet, Quanta, Wistron with its affiliate Wiwynn, and Inventec manufacture rack-scale systems for cloud providers, while Supermicro, Dell, HPE, Lenovo, ASUS and Gigabyte sell branded AI racks. Sanmina bought ZT Systems’ manufacturing business from AMD in Oct 2025; Foxconn’s Ingrasys is unlisted.
Open each company below for its source and limitations. Sector membership does not establish a confirmed supply contract.
Disclosed: AMD’s Advancing AI 2026 release describes Helios with 72 MI455X GPUs, 18 Venice CPUs and Pensando networking, and names its adopters and system partners.
Disclosed: Celestica’s Q3 2025 results exhibit and the Celestica–AMD release describe its switch programs, the rack-scale AI award and the Helios scale-up switches.
Disclosed: Cisco’s fiscal 2026 results give the hyperscaler AI orders and revenue; its newsroom describes Silicon One G300, the N9100 and UCS AI servers.
Disclosed: Dell’s 8-K gives AI-server orders, revenue and backlog; its blog and product pages describe the GB300 NVL72 systems, the IR7000 rack and the eRDHx.
Disclosed: Flex’s release describes its 800 VDC Power Rack for NVIDIA Vera Rubin, and its Sept 2026 8-K lists power shelves, PDUs, cold plates, manifolds and CDUs; NVIDIA names Flex as an 800 VDC partner.
Partner-disclosed: AMD’s release says HPE will be one of the first OEMs to offer its Helios rack-scale platform; NVIDIA names HPE among Blackwell Ultra server makers. HPE’s own GB200 NVL72 release was seen in search results only.
Disclosed: NVIDIA’s GB300 NVL72 and DGX GB300 pages give the GPU, CPU, NVLink, ConnectX-8, BlueField-3 and NVLink Switch counts, and its FY2026 10-K names its foundry, memory and assembly partners.
Disclosed: Jabil’s 10-Q and 10-K describe its AI-focused Intelligent Infrastructure segment, cloud data-center server platforms and the Hanley and Mikros acquisitions.
Evidence groupings are not investment ratings. Companies may appear in multiple sectors.
Also in this sector (not publicly listed)
Ingrasys Subsidiary of Hon Hai Precision Industry (2317.TW)
Taiwan-founded Foxconn group company offering GB200 NVL72 and GB300 NVL72 systems, NVIDIA MGX systems and rack-level liquid cooling; NVIDIA named it among Blackwell system makers (June 2, 2024). The holding entity inside the Hon Hai group is not disclosed in our sources; Hon Hai’s and FII’s rows reflect their own disclosures, not Ingrasys’s economics.
US hyperscale AI-rack manufacturer. AMD bought it in March 2025, kept the design operations (ZT Design Business) and sold the manufacturing business to Sanmina, closing Oct 27, 2025; ZT now operates inside Sanmina’s IMS segment, and its results are consolidated there.