A prominent venture capital investor has argued that two of Wall Street's most storied banking institutions, Goldman Sachs and JPMorgan Chase, are undergoing a quiet transformation that increasingly aligns their fortunes with the artificial intelligence sector. The claim, made during a Bloomberg "The Close" appearance by Momei Qu of PSP Growth, suggests that the connection between major banks and AI is now reflected in how these institutions talk about AI rather than in their reported earnings figures. The observation carries significant weight at a moment when the market is grappling with questions about how deeply AI capital cycles have become entangled with the traditional financial system.
The Core Thesis: Banks as AI Proxies
Qu's argument rests on a straightforward but provocative premise: the largest banks have inserted themselves into nearly every stage of the AI capital cycle. They provide debt financing for data center construction, they construct the bonds that fund massive infrastructure buildouts, they take AI companies public, and they advise on the mergers and acquisitions that reshape the competitive landscape. According to Qu, this positioning means the banks now function as leveraged bets on AI spending, even if their quarterly results still derive primarily from net interest income and balance sheet expansion.
The timing of this assertion is notable. It arrives amid a period of exceptional profitability for the banking sector. JPMorgan Chase, Citigroup, and Wells Fargo collectively reported a record $56.5 billion in net interest income for the quarter, a figure driven almost entirely by elevated interest rates and balance sheet growth rather than by any AI-related business line. That divergence between the narrative and the numbers is precisely what makes the thesis contentious. If banks were genuinely becoming AI stocks, one would expect their earnings reports to reflect AI-driven revenue streams. Instead, the current profit surge traces back to conventional macroeconomic factors.
This gap between perception and reported performance is not a minor technicality. It suggests that any "AI premium" embedded in bank valuations is a sentiment-driven phenomenon rather than an earnings-driven one. Investors may be pricing banks as AI beneficiaries in anticipation of future fee income from underwriting, advisory, and debt financing, but the present reality is that traditional banking dynamics continue to dominate the bottom line.
Market Skepticism and Divergent Signals
The market's behavior offers a counterpoint to the optimistic framing. Goldman Sachs shares have fallen sharply over the past three months, a decline that complicates the notion that investors are uniformly embracing the bank as an AI play. Citigroup, meanwhile, beat earnings expectations yet saw its stock drop 5.6 percent, marking the largest single-day decline since April 2025. These moves suggest that when investors actually price bank stocks, they are weighing share buyback slowdowns, capital adequacy ratios, and other conventional metrics far more heavily than any AI narrative.
The divergence extends across the banking landscape. Investment-banking-heavy institutions such as Goldman Sachs and Morgan Stanley stand to benefit disproportionately from a resurgence in AI-related initial public offerings and merger activity. Commercial and retail-focused banks like Wells Fargo, by contrast, remain anchored to interest rate dynamics and consumer credit conditions. Lumping all major banks together under a single "AI stock" label obscures these structural differences and risks overstating the breadth of the transformation.
Morgan Stanley has been explicitly identified as a likely beneficiary of a potential Anthropic IPO, a deal that would generate substantial underwriting fees for the lead banks involved. Goldman Sachs is similarly positioned. But the concentration of AI-related deal flow among a handful of elite investment banks raises its own set of questions about competitive dynamics and whether mid-tier institutions can capture meaningful share of this emerging revenue stream.
The Double-Edged Nature of AI Exposure
Perhaps the most important insight embedded in the argument is that the bank-AI linkage is not a one-way street. Just as banks share in the upside of AI infrastructure spending—through underwriting fees, debt financing interest, and advisory mandates—they also absorb the downside when AI valuations reset. This symmetry is often overlooked in bullish commentary.
Consider the credit exposure. Banks that finance data center construction carry direct balance sheet risk tied to the credit quality of those assets. If AI capital expenditure slows, the loans backing those projects could deteriorate. If technology cycles render certain computing assets obsolete faster than anticipated, the collateral value underpinning that debt could erode. This is not a hypothetical concern; it echoes the dynamics that made commercial real estate debt a source of systemic anxiety in previous cycles.
Then there is the fee income channel. Banks that build the bonds financing data center construction would see their fee income cut if AI spending decelerates. Banks that underwrite AI IPOs would see their pipelines dry up if market conditions sour. This is classic cyclical, transaction-driven revenue, not the kind of recurring, subscription-like income that commands premium valuations in stable markets.
Paul Tudor Jones has warned that an AI listing wave could make 2027 a difficult year for US equities. If that scenario materializes, the banks that feasted on AI deal flow during the boom would face a painful reversal. The comment attributed to the article's author that underwriting banks may "share in the impact of an AI valuation reassessment" captures this risk succinctly.
A New Financial-Real Economy Transmission Chain
Stepping back, the broader significance of the bank-AI coupling extends beyond the fate of individual institutions. It represents the emergence of a new transmission chain between the financial system and the real economy, one mediated by AI capital expenditure. AI infrastructure buildouts have become a major source of credit expansion, with debt financing flowing from bank balance sheets into data centers, chip fabrication facilities, and energy infrastructure. This means that a slowdown in AI spending would not merely hurt AI companies; it would ripple through the banks, bondholders, and structured product investors who financed the buildout.
The potential creation of new asset classes—data center asset-backed securities, collateralized loan obligations tied to AI infrastructure—introduces additional layers of complexity. The rating, pricing, and liquidity of these instruments would determine how smoothly the financial system absorbs an AI downturn. If these markets lack transparency or depth, the adjustment could be disorderly.
The parallel to commercial real estate and pre-2008 structured products is instructive but imperfect. AI infrastructure is a genuinely productive asset class with real demand drivers, unlike some of the speculative instruments that fueled the financial crisis. Nevertheless, the concentration risk inherent in a single sector driving a disproportionate share of debt issuance is a legitimate supervisory concern that has yet to receive the attention it deserves.
Commercialization Questions Remain Unanswered
From a commercialization standpoint, the AI industry's path to public markets is being charted primarily through capital market exits rather than product revenue growth. Anthropic's pre-IPO investor day, referenced in the original reporting, signals that leading AI companies are approaching the stage where public market validation becomes necessary. The observed dynamic in which other AI companies wait to see how the market prices the leader before proceeding with their own listings reflects a sequential game with significant herding effects.
This creates a binary outcome scenario. If Anthropic prices well, it could unlock a wave of AI listings that generates substantial fees for the underwriting banks. If Anthropic prices poorly, it could freeze the entire AI IPO window, leaving both the AI companies and their banking intermediaries stranded. The stakes for Goldman Sachs, Morgan Stanley, and their peers are therefore considerable.
What remains conspicuously absent from the available information is hard data on AI companies' unit economics. Gross margins, customer acquisition costs, net revenue retention, and cash flow profiles have not been thoroughly tested in public markets. Until they are, any valuation assigned to the AI sector—and by extension, any AI premium embedded in bank stocks—rests on assumptions rather than verified financial performance.
Infrastructure Financing and Its Hidden Risks
The infrastructure dimension of the story is where the financial and technological narratives intersect most directly. Data center construction is capital-intensive, long-duration, and dependent on assumptions about asset useful life, utilization rates, and residual value. When banks finance these projects, they are implicitly underwriting those assumptions. If technological change accelerates the obsolescence of computing infrastructure, or if power constraints limit the pace of buildout, the financing basis could prove shakier than anticipated.
The scale of AI capital expenditure appears to have reached systemic importance, given that it is large enough to move bank fee income structures. Estimates of total AI infrastructure debt remain difficult to pin down, but the magnitude implied by the reporting suggests hundreds of billions of dollars at minimum. This debt is deeply embedded in capital markets, creating linkages that would transmit stress quickly if AI spending decelerated.
Energy and power constraints represent a critical variable that the financial commentary largely omits. Data centers require enormous amounts of electricity, and the availability of power is emerging as a binding constraint on buildout timelines. This directly affects financing schedules, asset utilization rates, and ultimately the ability of borrowers to service their debt.
Competitive Dynamics and the Limits of the Thesis
The competitive landscape among banks themselves is another dimension that a blanket "banks are becoming AI stocks" framing obscures. Goldman Sachs and Morgan Stanley are positioned to capture a disproportionate share of AI IPO and M&A fees, potentially creating an oligopolistic dynamic in AI investment banking. JPMorgan's strength lies more in its diversified model and net interest income, which explains its robust guidance that exceeded market consensus.
Meanwhile, the rise of private credit and private equity as competitors to traditional banks in AI financing has received almost no attention. Non-bank lenders have been steadily encroaching on territory historically dominated by banks, and AI infrastructure financing could become a significant battleground. If private credit captures a meaningful share of this business, the AI upside for banks would be correspondingly diminished.
What to Watch
Several indicators will determine whether the "banks as AI stocks" thesis strengthens or fades. The pricing and market reception of Anthropic's IPO will serve as an immediate test of AI sector valuation and, by extension, the credibility of the AI-bank linkage. Quarterly earnings disclosures that break out AI-related business contributions would provide much-needed transparency. Cloud provider capital expenditure guidance serves as a proxy for AI buildout intensity and the associated financing demand.
On the risk side, data center debt defaults or rating downgrades would validate concerns about credit exposure. Progress by private credit and private equity in competing for AI financing mandates would signal erosion of the banks' competitive position. And the longer-term scenario articulated by Paul Tudor Jones—a 2027 AI listing wave that destabilizes equity markets—remains a tail risk worth monitoring over an 18-to-36-month horizon.
Assessing the Narrative
The proposition that Goldman Sachs and JPMorgan are becoming AI stocks is best understood as a narrative-driven, double-edged valuation claim rather than a statement of fundamental fact. The connection between bank fortunes and AI capital cycles is real and growing, operating through debt financing, IPO underwriting, and M&A advisory channels. But the same channels that deliver upside in a boom would amplify downside in a correction.
The argument's principal weaknesses are its reliance on a single primary source and its lack of quantified exposure data. The directional insight—that AI and finance are becoming more tightly coupled—is well-supported and consistent with observable market behavior. The specific prediction that banks should be valued as AI stocks is considerably less certain, especially when their current earnings continue to reflect traditional macroeconomic drivers rather than AI-related business activity. Investors would be wise to treat the AI premium in bank valuations with caution, recognizing that these institutions carry a dual beta exposure to both AI sentiment and financial sector fundamentals.
