But the company is unlikely to climb forever.
“The semiconductor sector has always been cyclical, so we should assume there will be ups and downs in the future too,” said Chris Miller, a history professor at The Fletcher School at Tufts University.
Miller, who authored The New York Times bestseller “Chip War: The Fight for the World’s Most Critical Technology,” told BI that chip giants like Nvidia and TSMC “have benefited immensely from the AI buildout.”
That, however, may not always be the case, he added.
“The boom in spending on AI infrastructure has been a major growth driver for the chip industry,” Miller said. “If excitement and investment in AI slow, chip industry growth will slow, too.”
What is the semiconductor supercycle?
At its core, a semiconductor supercycle simply means that there can be sustained periods of growth and increased demand — one that can last for several years.
But that rosy period of growth doesn’t last forever. The capital-intensive nature of semiconductor manufacturing, for one, makes it extremely difficult to align supply with demand.
And companies that want to meet a sudden uptick in demand must make hefty investments to boost production. While this would certainly raise supply levels, it also risks creating an oversupply if demand dies down.
This means that investors could be left hanging if they time the semiconductor supercycle wrongly.
For now, the chip industry has found itself buoyed by the surge in funding from AI-related investments.
That has resulted in a big payday for Nvidia, which has done brisk business with Meta’s Mark Zuckerberg and Tesla’s Elon Musk.
Zuckerberg even touted Meta’s growing stockpile of Nvidia chips in an interview with The Verge in January, adding that he expects the company to own more than 340,000 Nvidia H100 GPUs by the end of 2024.
However, some segments in the industry have already seen a slowdown, such as those relating to chips for consumer electronics. In April, chip manufacturer TSMC said it was lowering its growth forecasts, citing weak smartphone and personal computer sales.
The boom in data center builds has also spurred significant chip demand. If data center investment slows, chip firms could take a hit, Miller said.
To be sure, not everyone thinks we are at the peak of the AI-driven supercycle just yet.
“The rising demand for electricity due to digitalization, AI, and the use of renewable energies are increasing the demand for metal raw materials,” Thomas Rupf, chief investment officer at private bank VP Bank Asia, told BI.
“It is still too early to declare a new super cycle,” he said, adding that there are still investment opportunities in areas like industrial metals and mining, both boosted by AI demand.
Josie Ananto, a principal at EY-Parthenon’s strategy and transactions team, told BI that the growing ubiquity of chips means that demand will be a lot more persistent.
“The growth of the semiconductor sector is not a hype — it is real,” she said. “For example, electric vehicles in the automotive sector, product innovation in life sciences and pharmaceutical sector, consumer recommendation and personalization in the consumer products and retail sector.”
Heightened competition will only make things harder for incumbents like Nvidia
Semiconductor supercycle aside, investors may also want to pay attention to the competitive dynamics for chip designers like Nvidia.
“Nvidia today is the leader in producing chips for AI training and deployment,” Miller said. “However, Nvidia’s success is already attracting competition.”
Unlike TSMC, Nvidia designs but doesn’t make its own chips. This, Miller added, means that Nvidia is exposed to multiple threats — whether it be from Big Tech players like Meta and Google or startups like Groq.
“The barrier to entry for chip manufacturing is very high, given the significant capital investment required to open a fabrication plant,” EY-Parthenon’s Ananto said.
“For chip design, however, this is an area that is likely ripe for disruption, as more technology companies are investing in chip design capabilities, specifically tailored to their products with faster time to market,” she added.
Amazon, Meta, and Google are all designing their own AI chips.
And Nvidia isn’t standing still, either. In May, Nvidia CEO Jensen Huang said the company would release new chips on a “one-year rhythm” instead of its traditional two-year release cycle.
But that’s not to say that firms will all be competing to feast on the same pie. According to Miller, the pie may be huge now, but there’s no telling when it could shrink.
“The key question is how much Big Tech companies continue to spend building AI infrastructure,” he said.
According to a Goldman Sachs report published last month, tech companies are expected to pour over a trillion dollars into AI over the next few years.
However, the investment bank cautioned against being overly optimistic about AI’s return on investment, given how “exceptionally expensive” the technology is.
Read the full article here
