J.P. Morgan Warns of AI Market Euphoria

J.P. Morgan Warns of Investor Euphoria in AI Markets

J.P. Morgan warns that there are signs of investor euphoria in financial markets related to AI.

Since the launch of ChatGPT in 2022, only 42 AI companies in the S&P 500 have generated about 65 to 80% of the profits, revenues, and investments of the entire index, according to J.P. Morgan.

The rally in semiconductors shows technical patterns that closely resemble those of the internet bubble. Hedge funds are heavily invested in chip and hardware stocks, margin credit on the Korean stock market is rising, and retail traders are rushing into semiconductor options.

Leveraged chip ETFs, funds that amplify price fluctuations, have quintupled their influence on global stock markets since the beginning of 2024. Only a few stocks are driving the gains.

Stock Market Gains Concentrated in a Few Companies

J.P. Morgan also warns of extreme concentration. The ten largest U.S. stocks now account for about 40% of the S&P 500's market capitalization. In 2015, this figure was 17%.

The bank places these concentration figures in a global context. Despite the increase, the U.S. remains among the markets with relatively low stock concentration. Only India and Japan are less concentrated.

Nvidia still holds the largest share of the AI accelerator market, the specialized chips used for AI workloads, but this share is declining, dropping from 85% in 2023 to an estimated 75% by 2026, according to J.P. Morgan.

Custom chips from major cloud providers like Google's TPUs or Amazon's Trainium reduce operating costs by 30 to 40% compared to Nvidia GPUs. For example, Anthropic has committed to running its AI Claude on Amazon's Trainium for the next decade.

Four Warning Signs in the AI Market, According to J.P. Morgan

• Semiconductor stocks are deviating from their 200-day moving average as strongly as they did during the internet bubble.
• Hedge funds are more invested in chip stocks than ever before.
• Margin loans in Korea have tripled since 2020.
• Options trading on semiconductor stocks is five times higher than in 2020.

China Compresses Margins

J.P. Morgan also highlights familiar risks regarding the revenues of major AI labs like OpenAI and Anthropic. Their sales are growing rapidly, but computing costs are enormous, and future profitability remains uncertain.

Rising token prices could prompt companies to turn to cheaper open-source models. Signs of this shift are already appearing. Companies are moving tasks to less expensive models, average token prices are declining, and Chinese open-source models are approaching top-tier performance at a fraction of the cost.

The share of technology investments in economic growth is also rising, while the free cash flow margins of major cloud providers are shrinking and their debt financing is increasing.

Overall, J.P. Morgan asserts that AI creates multiple layers of concentration risk across markets, infrastructure, and the economy at large. NYU finance professor Aswath Damodaran has warned that the collapse of AI could hit harder than the bursting of the internet bubble.