The AI cabinet assembly supergroup debuts! What opportunities lie in this field?
If we compare the GPUs produced by NVIDIA to the core of computing power, but someone still needs to "stitch" them together, contract, and oversee the entire process to fit them into cabinets, wire them up, power them on, and add water cooling—adopting integrated solutions before they reach the customer's hands. This group of "assemblers" are the foremen of AI cabinet clusters!
1. First, understand two concepts: 🙋
ODM (Original Design Manufacturer): Does contract manufacturing for others + doesn't use its own brand, e.g., Taiwan's Hon Hai/Quanta, where the client finds and places orders with them.
OEM (Original Equipment Manufacturer): Own brand + sells complete products, e.g., DELL sells Dell PowerEdge.
Among the latter OEM path, a few have now stood out: $Dell Tech(DELL.US)$Super Micro Computer(SMCI.US)$Hewlett Packard Enterprise(HPE.US)$LENOVO GROUP(00992.HK) are clearly the unified supply platform for "second-tier AI clouds".
2. Who do they sell to? ❓
This is the key point to emphasize—the main customers served by these few OEMs are almost never the largest hyperscalers.
When the big four hyperscalers—Microsoft, Google, meta, and Amazon—build large-scale AI clusters, they basically bypass all OEMs and place orders directly with Taiwanese ODM factories.
Why is that?
(1) Volume is so large they don't need "branded solutions"
(2) They have their own complete SRE/engineering teams and don't need OEM services
(3) Ordering directly from factories can cut out the OEM brand premium
(4) Google even designs its own motherboards (Axion CPU + in-house ASIC)
So the real money DELL/HPE/SMCI /Lenovo make comes from customers "beyond the hyperscalers":
Second-tier AI clouds, Neocloud, transitioning crypto mining companies, sovereign wealth funds, governments, and traditional large enterprises.
3. This issue's stock list includes:
US Stocks Tier 1 $Dell Tech(DELL.US)$Super Micro Computer(SMCI.US)$Hewlett Packard Enterprise(HPE.US)
US Stocks Tier 2 $Penguin Solutions(PENG.US)$Cisco(CSCO.US)$IBM(IBM.US)$NetApp(NTAP.US)
HK Stocks NVIDIA Chain $LENOVO GROUP(00992.HK)
HK Stocks Domestic Substitution Chain $INSPUR DIGI ENT(00596.HK)$ZTE(00763.HK)
4. If we were to position each in one sentence, the differences among the few with high business purity are:
(1) SMCI is the Uniqlo of AI computing power (standardized, high volume, cheap, pure)
(2) DELL is the Carrefour of AI computing power (full range, wide coverage, strong brand)
(3) HPE is the Gome/Suning of AI computing power (traditional enterprise IT + government channels)
(4) Lenovo is the Xiaomi of AI computing power (cost-effective + China channels + overseas expansion)
(5) PENG is the luxury custom shop of AI computing power—high unit price, low volume, heavy service, many government clients, and its moat is "willing to do the dirty work others won't"
Longbridge community members, which one have you gotten on board with? Feel free to discuss below! 👏👏👏
AI rack assembly is the real star. If NVIDIA GPUs are the heart, someone still has to mount them into racks, connect cables, supply power, and add liquid cooling—these “assemblers” are the AI rack contractors. Today four giants lead the field: Dell, SMCI, HPE, and Lenovo, serving as the unified supply platform for second‑tier AI clouds. They are the OEMs—selling complete products under their own brands, like Dell’s PowerEdge. The pioneer in custom AI cluster integration is PENG, which focuses on research and government clients.