Goldman Sachs’ chief US equity strategist David Kostin has issued a contrarian warning about where the real AI bubble exists, and it’s not where most investors are looking. In a revealing interview on Goldman’s “Exchanges” podcast, Kostin argued that concerns about an AI-driven stock market bubble are misplaced when focused on publicly traded companies like Nvidia and other AI giants.
Instead, the true bubble is brewing in private AI markets, where capital availability and valuations have reached what Kostin describes as “probably unsustainable” levels. The retiring strategist, who spent 31 years at Goldman Sachs, drew on legendary investor George Soros’s theory of reflexivity to explain how private AI valuations are feeding on growth expectations rather than solid fundamentals. “As these firms are raising capital, the growth rate increases. As the growth rate increases, the valuation increases,” Kostin explained, highlighting a potentially dangerous feedback loop.
Kostin identified two critical risks in private AI markets: reflexivity and “circular financing” or “vendor financing,” where growth depends heavily on outside funding that may not be sustainable long-term. “At some point, the vendor doesn’t necessarily have the same growth to be able to fund that growth,” he warned.
In stark contrast, public AI markets show healthy fundamentals. Kostin cited Nvidia as a prime example: the AI chipmaker’s share price has increased 12-fold over the past three years, but critically, its earnings have also increased by the same magnitude. “So pretty much the price and the earnings have matched each other,” he noted, suggesting valuations are justified by actual performance.
Broader valuation metrics support this assessment. The largest companies in the S&P 500 index—many with significant AI exposure—currently trade around 30 times earnings. This is substantially below the 40 times peak seen in 2021 and far from the 50 times multiples during the dot-com bubble of the late 1990s.
Capital-raising trends further reinforce the distinction between public and private markets. The US has seen approximately 55 IPOs larger than $25 million this year—a fraction of the 280 in 2021 and nearly 400 in 1999. “There is capital availability in the public markets, but not necessarily ebullient. It’s there but not so dramatic,” Kostin concluded, suggesting public markets maintain pricing discipline that private markets currently lack.
Key Quotes
I believe in the private markets, the availability of capital, the price is probably unsustainable, which one could take as a synonym for a bubble
David Kostin, Goldman Sachs’ chief US equity strategist, made this statement on the firm’s podcast, directly identifying private AI markets as the location of bubble conditions rather than public markets. This represents a significant departure from mainstream concerns about publicly traded AI stocks.
As these firms are raising capital, the growth rate increases. As the growth rate increases, the valuation increases
Kostin explained the reflexivity dynamic in private AI markets, drawing on George Soros’s investment theory. This quote captures the potentially dangerous feedback loop where valuations become disconnected from underlying fundamentals, driven instead by capital availability and growth expectations.
So pretty much the price and the earnings have matched each other
Referring to Nvidia’s 12-fold increase in both share price and earnings over three years, Kostin used this observation to demonstrate that public AI markets are showing healthy fundamentals rather than bubble characteristics, contrasting sharply with private market dynamics.
There is capital availability in the public markets, but not necessarily ebullient. It’s there but not so dramatic
Kostin summarized the measured nature of public AI market activity, noting that IPO volumes remain well below historical bubble periods. This suggests public markets are maintaining pricing discipline and avoiding the exuberance that characterized previous technology bubbles.
Our Take
Kostin’s analysis offers a sophisticated framework for understanding AI investment risk that goes beyond simplistic bubble warnings. His distinction between public and private markets is particularly astute—public companies face daily price discovery and quarterly earnings scrutiny, creating natural valuation discipline. Private AI companies, however, operate in an environment where funding rounds can occur with limited transparency and where growth narratives can override fundamental analysis.
The timing of this warning is notable as it comes from a retiring strategist with no incentive to sugarcoat his views. The “circular financing” concern is especially relevant given reports of AI infrastructure companies funding each other’s growth. If Kostin is correct, we may see a private AI market correction that paradoxically strengthens public AI leaders like Nvidia, Microsoft, and Google, as capital flows back toward companies with proven revenue models and profitability. This could ultimately accelerate AI industry consolidation rather than derail AI development altogether.
Why This Matters
This analysis carries significant implications for the AI investment landscape as it challenges the prevailing narrative about where bubble risks actually exist. While media attention and investor anxiety have focused on the soaring valuations of publicly traded AI companies, Kostin’s warning about private markets highlights a potentially more dangerous dynamic unfolding away from regulatory oversight and daily price discovery.
The distinction matters because private market bubbles can have cascading effects when they eventually deflate. Companies that have raised capital at unsustainable valuations may struggle to meet growth expectations, leading to down rounds, layoffs, and reduced AI innovation. The “circular financing” pattern Kostin identifies is particularly concerning, as it suggests some AI startups’ growth may be artificially propped up rather than organically sustainable.
For the broader AI industry, this creates a bifurcated reality: public AI companies like Nvidia are demonstrating strong fundamentals with earnings matching valuations, while private AI ventures may be overheated. This could lead to a shakeout in the private AI sector without necessarily impacting the public AI giants, reshaping the competitive landscape and potentially consolidating AI development among established players with proven business models.
Recommended Reading
For those interested in learning more about artificial intelligence, machine learning, and effective AI communication, here are some excellent resources: