Key points
- Two chip CEOs on AI's next bottleneck
- When the chip shortages may ease
- What Hock Tan said about OpenAI
Broadcom (AVGO) CEO Hock Tan said Monday that access to capital could become the AI buildout's biggest constraint. He expects chip, memory, and power shortages to ease over the coming years. "[What] I think might be the biggest constraint is capital," Hock Tan said at the DBS Asia Future Forward Conference in Singapore, according to CNBC.
Intel (INTC) CEO Lip-Bu Tan, speaking at the same conference, pointed to the cost of capital and the pressure to show returns. "If there's any challenge in the front, it's capital and the cost of capital, and then the other part is in the longer run, people want to see the ROI," he said.
Hock Tan expects chip shortages to ease
Hock Tan said the industry has run into one supply bottleneck after another over the past two years and has overcome them. "Memory shortage today might not be that way a year from now," he said. He said wafer shortages could ease by 2028 or 2029, and power supply could improve as new infrastructure comes online.
He still sees demand running ahead of supply. "There is more demand for compute infrastructure than we can supply," Hock Tan said, adding that the gap could last through 2029 and possibly into 2030. He cautioned that demand could slow if businesses adopt advanced AI models more slowly than expected or if the models stop improving as quickly.
He also said Anthropic and OpenAI have shown they can generate significant revenue from their latest models, which are being used for AI personal assistants.
Lip-Bu Tan says the returns have to show up
Lip-Bu Tan said the industry is unlikely to see a significant slowdown over the next two to three years as more companies use AI and demand for computing power grows. After that, both executives said, growth will depend on how quickly businesses adopt AI and whether it delivers real economic benefits. Lip-Bu Tan said companies will increasingly need to show returns through efficiency gains and new business models.
Hock Tan defended OpenAI's valuation
Hock Tan rejected the idea that OpenAI is overvalued, according to The Standard in Hong Kong. He blamed investors for setting extreme expectations without understanding the technology, and he said many investors treat the sector like a casino.
The real test, he said, is whether frontier AI companies can build models that create enough value to justify the buildout. A day earlier, Wall Street Journal columnist Andy Kessler argued that falling AI prices could be what ends the data center boom.
AI borrowers face financing pressure
The 10-year Treasury yield was 5.24% on October 9. Bond investors are already asking for more yield to lend to the biggest AI spenders, Wellington Management's Loren Moran said in a Bloomberg segment on Sunday.
On October 8, the median issuer in the bond sample behind AIStockWire's AI Bubble Index, seven speculative-grade AI infrastructure borrowers plus Oracle, paid 334 basis points over Treasurys of similar maturity. The median bond in the SPDR Bloomberg High Yield Bond ETF (JNK) paid 231 basis points. The AI figure takes each issuer's median bond and then the median across issuers, while the JNK figure is the median across all of the fund's bonds.
Broadcom shares were down 0.5% at $359.66, and Intel was down 1.3% at $103.30 in premarket trading as of 8:20 a.m. ET.
