CEO CC Wei says TSMC is “trying to work very hard to narrow the gap” between supply and demand for AI chips
Taiwan Chip Giant TSMC reported profits jumping more than 61% year-on-year, driven primarily by demand for chips fueled by artificial intelligence and high performance computing (HPC).
Chairman and CEO CC Wei said he was trying to keep up with the massive demand for AI chips, and spoke about it in terms of narrowing down the gap between supply and demand rather than being able to fully meet demand.
“We're working very hard to narrow the gap for now,” Wei said in a call with TSMC investors.
TSMC's quarterly revenue rose nearly 44.4% year-on-year to over $30 billion, surpassing the company's guidance for the second quarter. We increased our third quarter guidance to $3.18-33 billion. This will increase by 38% year-on-year at the midpoint.
However, TSMC was affected by negative currency pressures, and executives also said they were conservative about the potential impacts of tariffs and trade restrictions, particularly in terms of the potential impact on consumer-related tips and the low price edge of the market.
CFO Jen-Chau Huang said in the call that high-performance computing (HPC) chips accounted for 60% of revenue during the second quarter, up 14%. Smartphone chips rose 7%, accounting for 27% of revenue, while IoT chips fell 14% as they accounted for just 5% of revenue. Automotive chips were flat at 5%, while TSMC's Digital Consumer Electronics (DCE) chips accounted for 1% of revenue, an increase of 30%.
TSMC's advanced processes (less than 7 nanometers) account for 74% of wafer revenues, while the 3 nm process accounts for 24% of wafer revenues. The company's capital expenditures at the time are expected to range from $38 billion to $420 billion.
“We believe that semiconductor demand is very basic and will continue to be robust,” Wei says. “Recent developments are also positive about the long-term demand outlook for AI. The explosive growth in token volumes indicates an increase in the use and adoption of AI models, which means more and more calculations are required, leading to cutting-edge silicon demand.
Work continues for TSMC Arizona's “Giga Fab Cluster”
TSMC says it plans to invest $165 billion in advanced semiconductor manufacturing in the US, with six wafer manufacturing facilities and two packaging fabs and research and development centers in Arizona.
“Our expansion plans allow TSMC to scale up to Giga Fab Cluster in Arizona to support cutting-edge customer needs in smartphones, AI and HPC applications,” says Wei.
TSMC is Nvidia's leading manufacturing partner. In a blog post in mid-April this year, Nvidia showed that it had commissioned Arizona more than a million square feet of manufacturing space to build and test Grace Blackwell Chips, and had asked Arizona to test its AI supercomputer in Texas. Nvidia said it plans to produce up to $50 trillion in AI infrastructure in the US through its partnership with TSMC, Foxconn, Wistron, Amkor and Spil.
TSMC's first Arizon Fab achieved mass production volume in the fourth quarter of last year with a 4 nm process. The second Fab offers 3 nm technology and is already complete, Wei said, and TSMC is working to “speed up its volume production schedule” to meet customer demand. Meanwhile, construction is underway on the third fab of the 2 nm technology, and TSMC hopes to accelerate its production schedule to meet AI chip demand. The fourth planned Fab is a 2 nm facility, with the fifth and sixth facilities expected to have even more advanced manufacturing processes.
TSMC has a new specialized fab in Japan, a second special fab in its work, and a planned auto chip-centric facility in Dresden, Germany. In Taiwan, the company plans to build more than a dozen new facilities over the next few years.
In addition to the production of AI chips, which are expected to help revolutionize the world, Wxecutives was also asked by Analyts how TSMC itself uses AI. Huang replied that the company uses AI in operations and manufacturing as well as R&D. For companies of the size of TSMC, Huang also pointed out that a 1% increase in productivity equals about $1 billion.
