March of this year $NVIDIA (NVDA.US)$ When introducing the Blackwell architecture chip, it may not have been anticipated that producing chips with this most powerful artificial intelligence (AI) chip architecture would encounter so many twists and turns. The latest revelation indicates that it has even led to internal conflicts between Nvidia and key manufacturing partners $Taiwan Semiconductor (TSM.US)$ in the process.

On Wednesday, October 16, Eastern Time, according to the tech media The Information citing insiders, NVIDIA began blaming Taiwan Semiconductor shortly after the release of Blackwell. Just a few weeks after the release, NVIDIA engineers found that the Blackwell architecture chips would not operate in high-pressure datacenter environments. Some NVIDIA engineers believe this may be partially due to design flaws in Blackwell.

Some NVIDIA engineers also believe that the slowdown in production of Blackwell chips is due to Taiwan Semiconductor adopting a new technology that packages different types of chips together. This explanation quickly began circulating among financial analysts following Taiwan Semiconductor and NVIDIA.

In addition to the design of Blackwell, two Taiwan Semiconductor employees stated that another factor from NVIDIA contributing to the problem was giving Taiwan Semiconductor less time to resolve issues compared to another major customer, Apple.

In early August of this year, the news of the delay in shipments of Blackwell chips shocked the market. Reports at the time suggested that due to design flaws, the most advanced AI chips in the Blackwell series would be delayed for three months or longer. The mass shipments of Blackwell chips might be delayed until the first quarter of next year. Morgan Stanley also stated in a report that production of Blackwell chips might pause for about two weeks but could catch up by the fourth quarter of this year through Taiwan Semiconductor’s efforts.

Following the news of the shipment delay, some anxious shareholders showed up on Wednesday. Taiwan Semiconductor's investor relations department shifted the blame onto NVIDIA.

The business dealings between NVIDIA and Taiwan Semiconductor began in 1995 and both sides can be said to have known each other in their early days. In that year, NVIDIA's founder and CEO, Huang Renxun, encountered a commercial bottleneck and wrote to Zhang Zhongmo, his senior at Stanford University and founder of Taiwan Semiconductor, for help. Not long after, Zhang Zhongmo personally called back Huang Renxun. Taiwan Semiconductor took on and excellently completed the order for NVIDIA's chip production, helping NVIDIA quickly take over the market. Huang Renxun turned this contact with Zhang Zhongmo into a comic strip and gave it to Zhang Zhongmo.

After that, Nvidia has a long-term cooperation with Taiwan Semiconductor, forming a highly symbiotic relationship. Nvidia relies on Taiwan Semiconductor's advanced technology and professional knowledge to implement chip design, and has become Taiwan Semiconductor's second largest customer after Apple. Analyst Dan Nystedt, who researches the semiconductor industry, estimated this year that Taiwan Semiconductor had $7.73 billion in revenue from Nvidia last year, accounting for 11% of the annual revenue, while Apple, the top contributor, contributed 25% of the revenue.

Strictly speaking, in the past nearly thirty years, Nvidia's cooperation with Taiwan Semiconductor has had its ups and downs. The news that broke out this Wednesday may not be very serious, and in the end, it may just be a footnote in the decades-long relationship between the two companies. Currently, Taiwan Semiconductor occupies over 90% of the advanced chip manufacturing market. Nvidia's launch of cutting-edge chips still depends on Taiwan Semiconductor, although there are signs that Nvidia hopes to reduce its dependence on Taiwan Semiconductor. For example, there are reports that Nvidia is researching collaboration with Samsung to produce new gaming chips, hoping to get a 20% to 30% discount on the technology to manufacture chips compared to Taiwan Semiconductor's chips of the same generation.

The report this Wednesday also mentioned another factor that maintains the cooperation between Nvidia and Taiwan Semiconductor, which is Morris Chang. Although he retired in 2018, he still has significant influence at Taiwan Semiconductor. He personally values long-term customer loyalty. Some customers have transferred some of the business originally given to Taiwan Semiconductor to other manufacturers. Later, when they tried to return to Taiwan Semiconductor, Morris Chang raised the price for them.

Whether or not it affects the cooperation between the two companies, the news on Wednesday underscores that as AI business attracts more and more funding, Nvidia is exerting increasing pressure on Taiwan Semiconductor. To consolidate its dominant position in the AI chip field, Nvidia is committed to introducing a new architecture every year. Taiwan Semiconductor has been striving to expand the capacity of its latest chip packaging technology, but in this respect, Taiwan Semiconductor's own efforts are limited.

Over three months ago, a report from the **** media, Mirror Weekly, exposed the contradictions in the development of the two companies under the AI boom. The media stated that in June this year, during a visit to Taiwan, Jensen Huang proposed to Taiwan Semiconductor that he hoped to set up a dedicated CoWoS production line outside the factory exclusively for Nvidia. Taiwan Semiconductor's executives retorted on the spot: "Is Nvidia willing to pay for it? Does Taiwan Semiconductor need to set up a wafer production line dedicated to Nvidia outside the factory?" The scene was very awkward as a result, but later Taiwan Semiconductor Chairman Wei Zhejia stepped in to defuse the tension.

Editor/Somer