Aswath Damodaran: Big Tech Has No Idea How AI Pays Off

SUMMARY

NYU finance professor Aswath Damodaran joins Scott Galloway and Ed Elson to analyze Big Tech's latest earnings and the implications of massive AI investments. The discussion covers the uncertainty around AI business models, the risks of overinvestment, and the challenges of valuing companies deeply tied to AI frontier labs.

MAIN POINTS

• Big Tech companies report strong revenue growth, but underlying shifts in business models raise questions for investors. [5:01]
• Concerns emerge over negative free cash flow and the lack of clear business models behind massive AI capital expenditures. [10:01]
• Apple stands out by avoiding the AI CapEx race, reflecting a more cautious approach compared to its peers. [15:02]
• Major AI revenue for Amazon, Microsoft, and Google is highly reliant on OpenAI and Anthropic, raising sustainability concerns. [18:01]
• Big Tech firms are seen as trying to recapture growth by investing heavily in AI, despite being mature, middle-aged companies. [23:01]
• Cross-holdings and intra-company investments complicate the valuation of Big Tech, with inflated earnings from paper gains in AI startups. [28:00]
• A shift is observed from concerns about compute supply to potential demand shortfalls, as AI infrastructure investments outpace end-user demand. [37:00]
• Chinese LLMs rapidly gain market share, intensifying competitive pressures and threatening to undercut Western AI business models. [39:01]
• Skepticism persists about whether OpenAI and Anthropic can achieve sustainable unit economics given high infrastructure costs. [44:03]
• Market anxiety about AI investments is present but not fully reflected in Big Tech valuations, with volatility and optimism coexisting. [47:01]
• Questions arise about off-balance-sheet debt and whether Big Tech is using its credibility to obscure financial risks. [52:01]
• Valuing AI frontier labs like OpenAI is fraught with uncertainty due to limited transparency and the need for more concrete business narratives. [56:01]
• Parallels are drawn between the current AI investment cycle and the dot-com bubble, with warnings of potential corrections ahead. [1:01:03]

DETAILED ANALYSIS

The conversation with Aswath Damodaran centers on the evolving landscape of Big Tech as companies like Microsoft, Amazon, Meta, and Google report robust revenue growth but face new challenges beneath the surface. While top-line numbers impress, Damodaran notes a fundamental shift: these firms are transitioning from asset-light, high-return models to capital-intensive operations, primarily driven by unprecedented investments in artificial intelligence infrastructure. Historically, these companies generated exceptional returns on invested capital with minimal physical assets, but the current AI race has transformed them into entities resembling manufacturing firms, now judged by their ability to generate returns on massive capital outlays.

A key concern is the lack of transparency and clarity regarding the business models underpinning these AI investments. Despite billions spent, none of the hyperscalers have articulated how they expect to monetize AI at scale. Damodaran argues that markets initially gave Big Tech the benefit of the doubt, assuming their track record would translate into successful AI ventures.

However, as capital expenditures balloon and free cash flows turn negative—Meta’s free cash flow dropped 91%, Amazon and Google’s went negative—investors are growing wary. The absence of clear narratives about whether these investments will lead to high-margin, premium products or low-margin, mass-market offerings leaves markets to speculate, fueling skepticism and volatility.

Apple emerges as a notable outlier, having opted out of the AI CapEx arms race. Its decision to wait out the uncertainty, rather than commit tens of billions to unproven AI infrastructure, reflects CEO Tim Cook’s cautious leadership. This approach contrasts sharply with peers who are aggressively building capacity without a defined end market.

Damodaran suggests that Apple’s strategy may prove prescient if the AI market bifurcates into premium and low-margin segments, especially as Chinese competitors aggressively target the latter with inexpensive large language models (LLMs).

Another layer of complexity arises from the intertwined relationships among Big Tech and AI startups. Wall Street estimates indicate that OpenAI and Anthropic account for over 70% of Amazon’s and Microsoft’s AI revenues, with similar patterns at Google. Much of this revenue is intra-company, stemming from AI startups purchasing compute from their Big Tech investors, rather than from true end-user demand.

Damodaran warns that this circular flow inflates reported revenues and earnings, masking the lack of genuine market traction. He emphasizes the need for greater transparency, advocating for separate reporting of AI divisions to clarify actual spending and returns.

Valuation challenges are compounded by cross-holdings and paper gains in AI startups. Amazon and Google’s net incomes have been significantly boosted by unrealized gains in Anthropic, OpenAI, and SpaceX, distorting traditional metrics like price-to-earnings ratios. Damodaran cautions against relying on net income or shortcut multiples, urging investors to focus on operating income and to adjust for the impact of non-operating investments.

The prevalence of intra-company financing and investing further muddies the waters, making it increasingly difficult to assess the true performance and risk profile of these conglomerates.

The discussion also addresses the risk of a supply-demand mismatch in AI infrastructure. Early narratives focused on compute scarcity, but recent developments—such as Meta and Musk attempting to sell excess compute capacity—suggest that demand may not be keeping pace with supply. This raises the specter of a demand crisis, with hundreds of billions in CapEx potentially chasing a market that is smaller or less profitable than anticipated.

The rapid rise of Chinese LLMs, which have captured over 50% market share in a matter of months, exacerbates competitive pressures and threatens to erode margins further, particularly in the mass-market segment.

Damodaran is skeptical that current AI business models can achieve sustainable unit economics, especially given the high fixed costs of data centers and specialized chips. He notes that, unlike traditional software, AI services incur significant marginal costs with each use, and scaling up does not necessarily lead to lower costs. While there may be opportunities for premium-priced, high-margin AI products, the bulk of the market is likely to be low-margin and highly competitive, challenging the viability of the current investment frenzy.

Despite these concerns, market valuations for Big Tech remain elevated, with anxiety about AI risks only intermittently reflected in stock prices. Damodaran attributes this to a combination of optimism, momentum, and the lack of a clear catalyst for a broader correction. He points out that the largest tech firms have manageable debt levels and could even benefit from a shakeout that eliminates weaker competitors.

However, he warns that off-balance-sheet debt and opaque financial structures could conceal additional risks, particularly for companies less able to weather a downturn.

Valuing AI frontier labs like OpenAI and Anthropic is particularly fraught, given the scarcity of financial disclosures and the reliance on speculative narratives. Damodaran advocates for constructing full, transparent stories—even if based on limited information—to prompt companies to provide more data and to foster more rigorous debate among investors. He draws parallels to the dot-com bubble, noting that while no two corrections are identical, the current cycle of hype, overinvestment, and shifting narratives bears a strong resemblance to the late 1990s.

The absence of major AI CapEx write-downs so far suggests that the market has yet to fully reckon with the risks, but Damodaran advises vigilance for signs of accounting revisions and restructuring charges as potential harbingers of a broader correction.

LINKS

• Prof G Markets newsletter subscription page.
• Order page for 'Notes On Being A Man'.
• Scott Galloway's Instagram profile.
• Ed Elson's Instagram profile.
• Ed Elson's X (Twitter) profile.
• Ed Elson's Substack newsletter.
• Prof G Markets on Spotify.
• Prof G Markets TikTok Q&A page.
• Prof G Markets homepage and resources.
• Claude AI platform, featured sponsor.
• VCX by Fundrise, public ticker for private tech.

KEYWORDS