AI

Big Tech Needs $600 Billion Annual Revenue to Justify AI Hardware Investments

07 July 2024

|

Zaker Adham

Summary

The technology sector is experiencing a surge driven by artificial intelligence, prompting major companies to invest heavily in building the necessary infrastructure to meet anticipated future demand. However, a recent analysis by David Cahn

, an analyst at Sequoia Capital, raises concerns about whether the revenue generated from AI will be sufficient to justify these significant expenditures.

In September, Cahn noted a substantial gap between the revenue expectations implied by AI infrastructure investments and the actual revenue growth within the AI sector. Initially, he estimated that $200 billion in annual AI revenue would be required to cover these investments. Almost a year later, with companies like Nvidia reaching unprecedented valuations, this figure has escalated to $600 billion annually.

Cahn’s calculations are based on the premise that for every dollar spent on a GPU, an equivalent amount is needed for the energy costs to run it in a data center. In Q4 2023, Nvidia's data center revenue forecast was $50 billion. By doubling this to account for total data center costs, and then doubling it again to reflect a 50% gross margin for end-users, he concluded that $200 billion in lifetime revenue is necessary to recoup the initial capital investment.

This figure excludes any margin for cloud vendors, implying an even higher total revenue requirement for a positive return on investment. The crucial question remains whether these infrastructure investments are driven by actual customer demand or merely by anticipated future needs.

Cahn predicts that the required AI revenue for investment payback will eventually reach $600 billion, particularly with Nvidia's forthcoming B100 chip, which promises 2.5 times better performance for only 25% more cost. This improvement is expected to spark a surge in demand for Nvidia chips.

Despite the significant expenditures, Cahn believes the investments will be worthwhile in the long run. He compares GPU capital expenditure to building railroads, suggesting that the demand will eventually materialize.

Executives from major tech companies have expressed confidence in AI's potential to drive revenue growth. For instance, Microsoft reported a notable increase in AI’s contribution to Azure’s growth. However, Cahn urges the industry to consider the broader implications, including who stands to benefit and who might suffer as these investments continue.

“There are always winners during periods of excess infrastructure building,” Cahn said. “Founders and company builders in AI will benefit from lower costs and the insights gained during this period of experimentation.”

Should his forecasts prove accurate, it could primarily be investors who bear the brunt of any fallout.