Financial
Could AI Speculation Trigger a Market Correction?
By John Abbott, CEO & Founder Emerge9
With over $1 trillion projected to be poured into AI over the coming years, it isn’t whether AI will transform the market but whether a speculative frenzy could destabilize it. As AI fever grips Wall Street, could this be the catalyst for the next major market correction?
A recent report from Goldman Sachs, Gen AI: Too Much Spend, Too Little Benefit, questions whether the expected benefits of AI could justify the massive levels of AI spending expected over the coming years. Specifically, Goldman’s Jim Covello states, “AI technology is exceptionally expensive, and to justify these costs, the technology must be able to solve complex problems, which it isn’t designed to do.”
However, the most recent earnings reports from the big tech companies suggest we are not in an AI bubble. We are starting to see tangible AI benefits at scale.
Based on its most recent 10Q, the big tech companies—namely Amazon, Meta, Microsoft, and Alphabet—are believed to account for 40% of Nvidia’s revenue. This spending has been incredibly consistent despite headlines that suggest otherwise.
The fundamentals for high ongoing investment in AI are excellent. During Microsoft’s Q2 2024 earnings call, CEO Satya Nadella stated that “over half of the Fortune 500 use Azure Open AI today.” Nadella also said Azure has over 53,000 Azure AI customers, a 50% increase over the last year. Moreover, Microsoft’s GitHub now has over 77,000 organizations using its Co-Pilot, which is up 180% year over year.
At the same time, Google is rolling out its Gemini AI tool, which is being integrated into its search engine experience. Given Google’s dominant 82% market share in search, Gemini can be considered critical to defending this position, particularly considering the threat posed by an AI-enabled Bing. Since Bing’s integration with Microsoft’s Copilot (Bing Chat), usage has attained new records, surpassing 140MM active users as of March 2024, up from 100MM active users over the prior 12 months.
On August 29th, Meta announced that companies, including Goldman Sachs and AT&T, use its Llama AI models for business functions like customer service, document review, and code generation. To date, 350 million users have downloaded the open-source Llama models, an increase of 50 million since the late July 2024 release of the Llama 3 model.
This deployment of AI-enabled tools across hundreds of millions of users signals a departure from the crypto/NFT bubble, which failed to produce consumer adoption beyond speculative use cases. However, we will continue to see questions about whether AI-linked productivity gains are commensurate with the massive levels of investment we’re seeing.
According to Anthropic CEO Dario Amodei, the cost of developing large language models (LLMs) is trending toward $10bn. Therefore, we can expect ownership of these models to be restricted to tech companies with market capitalizations in the hundreds of billions (or trillions) of dollars and nation-states. Although the capital-intensive and rapidly depreciating nature of GPU hardware could ultimately result in volatility that we have not previously seen in traditional software businesses, we should consider that the world’s deepest pockets are funding AI investments.