Nvidia Corp’s demand for advanced packaging from Taiwan Semiconductor Manufacturing Co (TSMC, 台積電) remains strong though the kind of technology it needs is changing, Nvidia CEO Jensen Huang (黃仁勳) said yesterday, after he was asked whether the company was cutting orders.
Nvidia’s most advanced artificial intelligence (AI) chip, Blackwell, consists of multiple chips glued together using a complex chip-on-wafer-on-substrate (CoWoS) advanced packaging technology offered by TSMC, Nvidia’s main contract chipmaker.
“As we move into Blackwell, we will use largely CoWoS-L. Of course, we’re still manufacturing Hopper, and Hopper will use CowoS-S. We will also transition the CoWoS-S capacity to CoWos-L,” Huang said on the sidelines of an event by chip testing and packaging company Siliconware Precision Industries Co (SPIL, 矽品精密) in Taichung.
Photo: An Rong Xu, Bloomberg
“So it’s not about reducing capacity. It’s actually increasing capacity into CoWoS-L,” he added.
Hopper refers to Nvidia’s graphics processing unit (GPU) architecture platform before the company announced Blackwell in March last year.
Nvidia has so far relied mainly on one type of CoWoS technology, CoWoS-S, to combine its AI chips.
TF International Securities Group Co (天風國際證券) analyst Kuo Ming-chi (郭明錤) on Wednesday said Nvidia was shifting its focus to a newer type of technology, CoWoS-L, and that suppliers would be affected.
In addition, local media reported that Nvidia was cutting CoWoS-S orders from TSMC in a potential hit to the Taiwanese chip foundry’s revenue.
Nvidia has been selling its Blackwell chips as quickly as TSMC can make them, but packaging has remained a bottleneck due to capacity constraints.
Still, Huang said that the amount of advanced chip packaging capacity was “probably four times” the amount available less than two years ago.
He declined to answer questions on the new US export restrictions that limit AI chip exports to most countries except for a select group of close US allies including Taiwan.
Huang was in Taichung yesterday to attend the opening ceremony of a new SPIL plant in the city’s Tanzi Technology Industrial Park (潭子科技產業園區).
“The technology that we’re working on is becoming more sophisticated. Chips are getting more and more complex, and the packaging technology will need to evolve as well. What’s even more exciting is the integration of silicon photonics, enabling us to connect multiple packages into one massive system,” Huang said.
He said Nvidia’s partnership with SPIL would be instrumental in pushing the boundaries of innovation in coming years.
Huang also addressed the broader implications of AI development, saying that AI combined with robotics will bring tremendous benefits to Taiwan’s world-leading electronics industry.
SPIL expressed enthusiasm for the collaboration, stating that Huang’s visit highlights the strong relationship between the two companies.
Industrial sources told CNA that Huang’s visit to SPIL aims to secure production of 3D CoWoS packaging services used in AI chip manufacturing.
TSMC, which has entered advanced CoWoS development, has outsourced the services to SPIL to meet strong global demand, the sources said.
As its CoWoS services have secured certification, SPIL is expected to raise production gradually starting from the second quarter of this year, the sources added.
It was the first public event for Huang in Taiwan this year. He is also expected to attend Nvidia Taiwan’s annual Lunar New Year party this week in Taipei.
In addition, Huang is to meet with Taiwanese tech executives from Hon Hai Precision Industry Co (鴻海), Quanta Computer Inc (廣達) and Inventec Corp (英業達), which all roll out AI servers powered by Nvidia’s chips, as well as PC brands Asustek Computer Inc (華碩) and Micro-Star International Co (微星).
Huang, who was born in Tainan, Taiwan’s historic capital, before emigrating to the US at the age of nine, is hugely popular in the country.
Additional reporting by CNA and AP
Taiwan’s foreign exchange reserves fell below the US$600 billion mark at the end of last month, with the central bank reporting a total of US$596.89 billion — a decline of US$8.6 billion from February — ending a three-month streak of increases. The central bank attributed the drop to a combination of factors such as outflows by foreign institutional investors, currency fluctuations and its own market interventions. “The large-scale outflows disrupted the balance of supply and demand in the foreign exchange market, prompting the central bank to intervene repeatedly by selling US dollars to stabilize the local currency,” Department of Foreign
ENERGY ISSUES: The TSIA urged the government to increase natural gas and helium reserves to reduce the impact of the Middle East war on semiconductor supply stability Chip testing and packaging service provider ASE Technology Holding Co (日月光投控) yesterday said it planned to invest more than NT$100 billion (US$3.15 billion) in building a new advanced chip testing facility in Kaohsiung to keep up with customer demand driven by the artificial intelligence (AI) boom. That would be included in the company’s capital expenditure budget next year, ASE said. There is also room to raise this year’s capital spending budget from a record-high US$7 billion estimated three months ago, it added. ASE would have six factories under construction this year, another record-breaking number, ASE chief operating officer Tien Wu
The EU and US are nearing an agreement to coordinate on producing and securing critical minerals, part of a push to break reliance on Chinese supplies. The potential deal would create incentives, such as minimum prices, that could advantage non-Chinese suppliers, according to a draft of an “action plan” seen by Bloomberg. The EU and US would also cooperate on standards, investments and joint projects, as well as coordinate on any supply disruptions by countries like China. The two sides are additionally seeking other “like-minded partners” to join a multicountry accord to help create these new critical mineral supply chains, which feed into
For weeks now, the global tech industry has been waiting for a major artificial intelligence (AI) launch from DeepSeek (深度求索), seen as a benchmark for China’s progress in the fast-moving field. More than a year has passed since the start-up put Chinese AI on the map in early last year with a low-cost chatbot that performed at a similar level to US rivals. However, despite reports and rumors about its imminent release, DeepSeek’s next-generation “V4” model is nowhere in sight. Speculation is also swirling over the geopolitical implications of which computer chips were chosen to train and power the new
RSS