Longwell enters Nvidia supply chain
What happened
- Longwell said on May 25 it entered Nvidia’s next-generation AI server supply chain after its high-power cables passed platform certification, DigiTimes reported Tuesday. - The key detail is pilot production: Longwell said the certified cables have moved into market qualification and pilot runs for Nvidia’s next platform. - Longwell is accelerating Thailand expansion, where its existing factory and new capital plan support AI and low-earth-orbit demand.
Why it matters
Longwell has joined Nvidia’s supply chain for a next-generation AI server platform after its high-power cables passed certification and moved into qualification and pilot production, according to DigiTimes and Longwell’s own statement. The Taiwan-listed cable and connector supplier said the update on May 25. The development adds another component maker to the hardware stack behind Nvidia’s AI systems at a time when power delivery and interconnects are becoming more critical inside denser server racks. Nvidia has also been pushing a higher-voltage data-center power architecture for future AI factories, underscoring why cable and connector vendors matter in this build-out. ### What exactly did Longwell say it won? Longwell said its high-power cables cleared “the latest chip platform certification” and had entered market qualification and pilot production for Nvidia’s next-generation AI server platform, according to the DigiTimes report. That wording matters because it places the company beyond a design pitch and into the early manufacturing phase suppliers typically need to pass before volume shipments. (digitimes.com) DigiTimes identified Longwell as a cable and connector supplier, a category that sits lower in the bill of materials than GPUs or server chassis but is directly exposed to changes in rack power density and system design. Longwell’s website says it supplies cables and electronic components to global electronics manufacturers. ### Why are high-power cables getting more attention in AI servers? (digitimes.com) Nvidia said in a May 2025 technical blog that traditional 54-volt direct-current power distribution is running into limits as AI racks move toward megawatt-scale power loads. The company said its 800-volt HVDC architecture is intended to support 1 megawatt IT racks and beyond starting in 2027. That shift puts more weight on upstream suppliers that can meet power, heat and certification requirements for denser systems. DigiTimes separately reported this week that AI server suppliers were contending less with order demand than with shortages in power and other key resources, reinforcing how power-related components have become a constraint in the supply chain. (developer.nvidia.com) ### What does Thailand have to do with this? Longwell has been building manufacturing capacity in Thailand for several years. The company said in a 2020 post that it established a factory in Chachoengsao, east of Bangkok, for capacity expansion and to avoid additional U.S. tariffs, with initial power-cord lines and monthly output of 250,000 pieces. On May 8, Longwell’s board approved a 600 million baht capital increase for its Thailand subsidiary, according to a Taiwan market filing carried by financial data services. (digitimes.com) DigiTimes said the company is now speeding Thailand expansion to serve AI and low-earth-orbit, or LEO, demand. ### Where does this fit in Nvidia’s broader supplier web? DigiTimes’ report points to a supply chain that is still widening below the headline chip makers and server assemblers. (longwell.com) New entrants at the cable level suggest Nvidia’s next platforms are pulling in more specialized suppliers as power and interconnect requirements rise. That is an inference from Longwell’s certification milestone and Nvidia’s own published power-roadmap direction. (finance.biggo.com) The next markers to watch are whether Longwell moves from pilot production to volume shipments and whether Nvidia names more partners tied to its future rack-power architecture. Nvidia has said its 800-volt HVDC approach is aimed at deployments starting in 2027, giving suppliers a defined timetable for the next phase of qualification and build-out. (developer.nvidia.com) (digitimes.com)
Key numbers
- Longwell said on May 25 it entered Nvidia’s next-generation AI server supply chain after its high-power cables passed platform certification, DigiTimes reported Tuesday.
- The Taiwan-listed cable and connector supplier said the update on May 25.
- (digitimes.com) Nvidia said in a May 2025 technical blog that traditional 54-volt direct-current power distribution is running into limits as AI racks move toward megawatt-scale power loads.
- The company said its 800-volt HVDC architecture is intended to support 1 megawatt IT racks and beyond starting in 2027.
What happens next
- Longwell has joined Nvidia’s supply chain for a next-generation AI server platform after its high-power cables passed certification and moved into qualification and pilot production, according to DigiTimes and Longwell’s own statement.
- The Taiwan-listed cable and connector supplier said the update on May 25.
- Longwell said its high-power cables cleared “the latest chip platform certification” and had entered market qualification and pilot production for Nvidia’s next-generation AI server platform, according to the DigiTimes report.
Quick answers
What happened in Longwell enters Nvidia supply chain?
Longwell said on May 25 it entered Nvidia’s next-generation AI server supply chain after its high-power cables passed platform certification, DigiTimes reported Tuesday. The key detail is pilot production: Longwell said the certified cables have moved into market qualification and pilot runs for Nvidia’s next platform. Longwell is accelerating Thailand expansion, where its existing factory and new capital plan support AI and low-earth-orbit demand.
Why does Longwell enters Nvidia supply chain matter?
Longwell has joined Nvidia’s supply chain for a next-generation AI server platform after its high-power cables passed certification and moved into qualification and pilot production, according to DigiTimes and Longwell’s own statement. The Taiwan-listed cable and connector supplier said the update on May 25. The development adds another component maker to the hardware stack behind Nvidia’s AI systems at a time when power delivery and interconnects are becoming more critical inside denser server racks. Nvidia has also been pushing a higher-voltage data-center power architecture for future AI factories, underscoring why cable and connector vendors matter in this build-out. What exactly did Longwell say it won? Longwell said its high-power cables cleared “the latest chip platform certification” and had entered market qualification and pilot production for Nvidia’s next-generation AI server platform, according to the DigiTimes report. That wording matters because it places the company beyond a design pitch and into the early manufacturing phase suppliers typically need to pass before volume shipments. (digitimes.com) DigiTimes identified Longwell as a cable and connector supplier, a category that sits lower in the bill of materials than GPUs or server chassis but is directly exposed to changes in rack power density and system design. Longwell’s website says it supplies cables and electronic components to global electronics manufacturers. Why are high-power cables getting more attention in AI servers? (digitimes.com) Nvidia said in a May 2025 technical blog that traditional 54-volt direct-current power distribution is running into limits as AI racks move toward megawatt-scale power loads. The company said its 800-volt HVDC architecture is intended to support 1 megawatt IT racks and beyond starting in 2027. That shift puts more weight on upstream suppliers that can meet power, heat and certification requirements for denser systems. DigiTimes separately reported this week that AI server suppliers were contending less with order demand than with shortages in power and other key resources, reinforcing how power-related components have become a constraint in the supply chain. (developer.nvidia.com) What does Thailand have to do with this? Longwell has been building manufacturing capacity in Thailand for several years. The company said in a 2020 post that it established a factory in Chachoengsao, east of Bangkok, for capacity expansion and to avoid additional U.S. tariffs, with initial power-cord lines and monthly output of 250,000 pieces. On May 8, Longwell’s board approved a 600 million baht capital increase for its Thailand subsidiary, according to a Taiwan market filing carried by financial data services. (digitimes.com) DigiTimes said the company is now speeding Thailand expansion to serve AI and low-earth-orbit, or LEO, demand. Where does this fit in Nvidia’s broader supplier web? DigiTimes’ report points to a supply chain that is still widening below the headline chip makers and server assemblers. (longwell.com) New entrants at the cable level suggest Nvidia’s next platforms are pulling in more specialized suppliers as power and interconnect requirements rise. That is an inference from Longwell’s certification milestone and Nvidia’s own published power-roadmap direction. (finance.biggo.com) The next markers to watch are whether Longwell moves from pilot production to volume shipments and whether Nvidia names more partners tied to its future rack-power architecture. Nvidia has said its 800-volt HVDC approach is aimed at deployments starting in 2027, giving suppliers a defined timetable for the next phase of qualification and build-out. (developer.nvidia.com) (digitimes.com)
