Nvidia relies on TSMC to supply customers with computing accelerators not only for wafer processing but also for packaging chips using its unique CoWoS spatial packaging method. The former’s attempts to secure dedicated TSMC manufacturing capacity for these needs have reportedly failed.
Image source: Nvidia
This information is published by the resource Mirror Mediareferring to details of the program of the visit of Nvidia founder Jensen Huang to Taiwan in June of this year. According to sources, he met then not only with the retired founder of TSMC Morris Chang, but also with the current head of the company CC Wei. The head of Nvidia at that time, if rumors are to be believed, asked TSMC to allocate a separate production line for the needs of his company, on which AI chips of this brand would be packaged, but was refused by representatives of the Taiwanese contractor. The outcome of the negotiations created some tension in the relations between the companies, as noted by sources, but the current chairman of the board of directors CC Wei did everything possible to smooth over the consequences.
At a recent quarterly conference call, TSMC management admitted that the company would not be able to meet demand for the production of components for artificial intelligence systems until at least 2026. At the same time, the Taiwanese manufacturer still cannot find a balance between supply and demand, but is trying to rationally determine the amount of necessary capital expenditures. Apparently, investments in a production line for packaging Nvidia chips at this stage seem irrational to TSMC management. Moreover, the profit margin in this service area is approaching the company’s average, without providing any impressive advantages. As noted, TSMC’s refusal was motivated by possible consequences for the company’s relations with other clients who would also like to achieve certain privileges. By maintaining equal conditions for all clients, TSMC can ensure a more predictable situation with scaling production capacities.
In the past, as sources familiar with TSMC’s practice note, the company has provided certain privileges to large customers. For example, Apple once asked to provide it with dedicated chip production lines, and TSMC agreed to this, but at that time the Taiwanese manufacturer was heavily dependent on Apple’s orders and could not ignore such an opportunity to optimally load its conveyor. In the case of Nvidia, the situation is noticeably different. TSMC is not expected to be able to cover the market’s needs for chip packaging capacities using the CoWoS method even by the end of next year, since demand will grow at an accelerated rate. Such a position in the market, close to a monopoly, allows TSMC to more firmly defend its interests in negotiations with customers. This is noticeable even in the statements of C.C. Wei, who recently admitted that he would like to take more money from the same Nvidia for TSMC’s services.
If you notice an error, select it with your mouse and press CTRL+ENTER.
