GenAI Must Generate $600B A Year To Justify Hardware Costs, Sequoia Estimates: How Much Does Google Search Make?

Artificial intelligence (AI) has taken the markets by storm, catapulting several AI-related stocks to all-time highs.

AI’s rally is a “bubble,” concerned investors note. And a new report from Sequoia Capital cast doubt on whether firms will ever be able to recoup hardware costs.

Sequoia’s Report: A report from Sequoia Capital written by David Cahn estimated that AI needs to generate $600 billion a year to justify its immense hardware costs.

Cahn arrived at the $600 billion number by taking NVIDIA Corp‘s NVDA fourth-quarter GPU revenue of $150 billion and doubling it to account for the cost of AI data centers. Cahn doubled the resulting $300 billion to account for a 50% gross margin for the end-user of the GPU.

GPU stockpiles have steadily increased as supply shortages have diminished.

Cahn assumes that Alphabet Inc, Microsoft Corp, Apple Inc, Meta Platforms Inc, Oracle Corp, ByteDance, Alibaba Group, Tencent Holdings, X and Tesla Inc will generate a combined $100 billion from new AI-related revenue each year. The estimate creates a $500 billion gap between revenue and cost.

What About Google Search?: Alphabet Inc’s flagship Google Search made $175 billion in 2023, according to company filings. Google is the world’s preeminent search engine and captures a gigantic market share.

Therefore, generative AI must generate nearly 3.5 times the annual revenue of Google Search to turn a profit, a tall task indeed.

Market Implications: While some signs are encouraging that AI will eventually prove valuable to consumers, it is no sure thing that it will ever turn a profit over GPU stockpile costs.

In determining the industry’s prospects, experts (such as Forbes’ Peter Cohan) have noted the importance of an “AI killer,” a use case that is a clear, highly desirable value-add for consumers. A historical example is spreadsheets during the personal computing revolution.

No such feature exists yet, as customers currently seem reluctant to spend money on AI. It may be months or years before a killer use case exists, if ever. The release of OpenAI’s long-awaited GPT-5 could serve as a potential catalyst if the release lives up to lofty expectations. Alternatively, AI’s future use case may be far from our current expectations.

Now Read:

• Despite The Market Hype, Only 5% Of US Firms Use Generative AI In Production

Image: Pixabay