What Foxconn (Hon Hai) makes for data centers
public · HQ New Taipei City, Taiwan · honhai.com · also known as: Ingrasys
Foxconn, operating globally as Hon Hai Technology Group, is a contract manufacturer (ODM) that builds Nvidia and AMD rack-scale AI systems, including NVL72-class cabinets, GPU baseboards and compute trays, which it builds in-house through its Ingrasys subsidiary. It also makes board connectors and copper interconnect cables through its Foxconn Interconnect Technology (FIT) unit, server fans, chassis and rack enclosures, direct-to-chip cold plates, and coolant distribution units.
Parts, by layer
Layer 4 — Servers, racks & integration
| Part | Role | Confidence | Competitors | Note |
|---|---|---|---|---|
| Rack-scale systems | odm | inferred | Quanta Computer, AMD, ASRock Rack, ASUS, Cisco, Compal | Ingrasys |
| EMS, system integration & rack-level test | manufacturer | inferred | Quanta Computer, Benchmark, Celestica, Compal, Flex, Inventec | US and Mexico rack-integration sites run by |
| GPU baseboards & compute trays | manufacturer | inferred | Quanta Computer, Inventec, Nvidia, Pegatron, Supermicro, Wistron | boards built by |
| Fans | manufacturer | inferred | AVC, Delta Electronics, Nidec, Sanyo Denki, Sunon, Cooler Master | |
| Chassis, rails, rack enclosures | manufacturer | inferred | Chenbro, King Slide, Wistron, AIC | |
| Air-cooled heatsinks & vapor chambers | manufacturer | inferred | AVC, Auras, Chaun-Choung CCI, Delta Electronics, Jentech, Boyd |
Layer 5 — Liquid cooling (IT-side: chip to CDU)
| Part | Role | Confidence | Competitors | Note |
|---|---|---|---|---|
| Direct-to-chip cold plates | in house | inferred | AVC, Auras, Delta Electronics, Flex, Fositek, Jentech | in-house for their own racks |
| Rack manifolds | manufacturer | inferred | AVC, Auras, Delta Electronics, Kaori, Parker Hannifin, Vertiv | |
| Coolant distribution units | manufacturer | inferred | Auras, Daikin, Delta Electronics, Envicool, Johnson Controls, Kaori |
Where Foxconn sits in the stack
Foxconn's placements in this catalog cover 9 parts, from layer 4 (servers, racks & integration) to layer 5 (liquid cooling (it-side: chip to cdu)). None of them currently carries a sourced supply constraint tag. Across these parts it overlaps with other companies most often on the lines below.
What these parts are
- Rack-scale systems — A rack-scale system is the fully integrated cabinet of compute trays, switch trays, power shelves, cooling manifolds and busbar that ships as one wired unit rather than as discrete servers.
- EMS, system integration & rack-level test — Electronics manufacturing services firms assemble, wire and burn-in the finished server or rack before it ships, standing between the ODM board build and the customer's data center floor.
- GPU baseboards & compute trays — The baseboard is the printed circuit assembly that carries the GPUs (and often paired CPUs) and their direct power and interconnect, packaged into a tray that slides into the rack.
- Fans — The axial and centrifugal fans mounted in server chassis and rack trays that move air across heatsinks and boards, still needed even in racks that use liquid cooling for the highest-power chips.
- Chassis, rails, rack enclosures — The sheet-metal chassis and mounting rails that trays slide into, and the enclosure (the rack cabinet itself) that houses the whole assembly.
- and 4 more in the table above.
Closest peers in this catalog
Seen in real machines
Foxconn (Hon Hai) is named in 1 of the rack teardowns and supercomputer breakdowns on this site. Each line below is a piece of that machine.
- Nvidia GB300 NVL72 · rack teardown
- CDU (in-rack or in-row)
Monthly revenue (TWSE/TPEx filing)
| Month | Revenue (TWD m) | MoM | YoY |
|---|---|---|---|
| 2026-09 | 1,158,633 | 25.7% | 38.4% |
| 2026-08 | 921,766 | -2.6% | 52.0% |
| 2026-07 | 946,513 | 15.2% | 54.2% |
| 2026-06 | 821,763 | -4.4% | 52.1% |
| 2026-05 | 859,409 | 3.3% | 39.6% |
| 2026-04 | 832,098 | 3.5% | 29.7% |
| 2026-03 | 803,738 | 34.9% | 45.6% |
| 2026-02 | 595,813 | -18.4% | 8.1% |
| 2026-01 | 730,039 | -15.4% | 35.5% |
| 2025-12 | 862,861 | 2.2% | 31.8% |
| 2025-11 | 844,276 | -5.7% | 25.5% |
| 2025-10 | 895,709 | 7.0% | 11.3% |
From the blog
Latest feed
- Taiwan raising $4 billion to build industrial park in Texas to support US chipmaking supply chain — Tom's Hardware, 2026-10-16Taiwan plans to build a $4 billion industrial park in Texas to help smaller Taiwanese companies expand their presence supporting major chipmakers.
- Taisic Materials expands SiC capacity for AI data centers, 8-inch markets — DigiTimes, 2026-10-13Taisic Materials is expanding its Silicon Carbide (SiC) production capacity to serve the power electronics and electrical equipment needs of AI data centers.
- Robots still can't build Nvidia's AI servers as well as people can — DigiTimes, 2026-10-07The article discusses the limitations of using robotics to assemble complex Nvidia AI server hardware compared to human manual assembly.
- TSMC, ASE break ground on Taiwan's first advanced packaging cluster in Kaohsiung — DigiTimes, 2026-09-22Taiwan is establishing its first advanced packaging cluster in Kaohsiung, led by TSMC, in response to rising demand from AI and HPC.
- Tesla AI server order puts Pegatron back alongside Asus, 18 years after split — DigiTimes, 2026-09-21Asus and Pegatron are collaborating on AI server designs, aiming to compete with established players like Quanta, Wistron, Foxconn, and Gigabyte.
- Foxconn Interconnect Technology targets the scaling challenge in AI optical interconnects — DigiTimes, 2026-09-17Foxconn Interconnect Technology held an event themed 'Light at Scale' to discuss the next generation of AI connectivity, covering optics, photonics, and manufacturing challenges.
- Foxconn links FII and FIT to tackle AI infrastructure beyond the chip — DigiTimes, 2026-09-17Foxconn is advancing beyond just supplying chips, offering comprehensive system-level solutions that integrate AI servers, high-speed interconnects, optical communications, power, and cooling.
- Sharp opens orders for Foxconn-built Nvidia AI servers, targeting JPY250 billion by fiscal 2030 — DigiTimes, 2026-09-16Sharp Corporation began taking orders for Nvidia AI servers, manufactured by Foxconn, aiming to generate JPY250 billion in revenue by fiscal 2030.
- Foxconn subsidiary broadens AI interconnect role as 1.6T enters deployment — DigiTimes, 2026-09-14AI data centers are rapidly adopting 1.6T connectivity, while 800G becomes standard for AI clusters and 400G remains prevalent in general enterprise use.
- AI infrastructure race shifts to Grid-to-Chip as Foxconn, Delta, Flex converge on the same track — DigiTimes, 2026-09-14Industry focus for AI infrastructure is shifting from compute components to power, thermal management, and electrical grid components, exemplified by Flex's acquisition of EPC Power.
Questions about Foxconn (Hon Hai)
- What does Foxconn (Hon Hai) make for AI data centers?
- This catalog places Foxconn (Hon Hai) on 9 parts: Rack-scale systems (odm), EMS, system integration & rack-level test (manufacturer), GPU baseboards & compute trays (manufacturer), Fans (manufacturer), Chassis, rails, rack enclosures (manufacturer), Air-cooled heatsinks & vapor chambers (manufacturer), Direct-to-chip cold plates (in house) and Rack manifolds (manufacturer), and 1 more.
- Is Foxconn (Hon Hai) publicly traded?
- Yes — Foxconn (Hon Hai) is listed on Taiwan Stock Exchange under 2317.TW. Prices shown on this page are a nightly snapshot, not real-time, and nothing here is investment advice.
- Where is Foxconn (Hon Hai) based?
- New Taipei City, Taiwan, per the company record in this catalog.
- Who else makes rack-scale systems for AI data centers?
- Quanta Computer, AMD, ASRock Rack, ASUS, Cisco and Compal are also placed on Rack-scale systems in this catalog.
- How well sourced are these placements?
- Every part→company placement carries a confidence tag. For Foxconn (Hon Hai) the mix is 9 inferred. "disclosed" means the company or a filing says so; "reported" comes from trade press or a reference-design list; "inferred" is a lead from the taxonomy with no citation yet. Treat reported and inferred as leads, not facts.
Answers are generated from this catalog's own rows — see sources & method.
Verification: enriched, last verified 2026-08-27. Sources: www.honhai.com · www.honhai.com.· Markdown · JSON