AI fears are pushing S&P Global below fair value
S&P Global looks mispriced, not broken. The stock is 22% below our $570 fair value after a selloff driven by fears that AI models built for financial research could disrupt capital markets information services. That sets up the core debate: is S&P an infrastructure layer for finance, or just another automatable data vendor?
The market is starting to price it like the latter. 72% of S&P 500 companies disclosed at least one material AI risk in 2025, up from 12% in 2023, with financials among the most exposed. Bears read that surge as a sign that trusted data workflows could be bypassed faster than most investors expect.
Still, that view looks too broad for S&P Global. We keep our $570 fair value estimate because the pressure appears concentrated in Market Intelligence, while Ratings, Indices, and Energy look less vulnerable to AI disruption. That distinction matters: in the benchmark and ratings businesses, the moat rests on intangible assets and network effects, with acceptance among index users and regulators helping sustain the economics.
If the market is front-running a broad AI reset that has not yet shown up in the durable profit base, the setup looks more asymmetric than merely cheap. The thesis weakens if those benchmark businesses stop behaving like infrastructure.
Market Intelligence is the most plausible AI pressure point
The market is not wrong to worry about AI across every corner of S&P Global. It is wrong to assume the economics change everywhere at once.
Where AI can actually bite
The plausible bear case starts in Market Intelligence. That is the segment most exposed to self-service analytics, automated research workflows, and AI agents pulling summaries, comparables, or research notes on demand. Recent product conversation around models built for financial research makes that fear feel immediate, at least for some point products.
But the real question is not whether AI can demo a shortcut. It is whether clients abandon trusted inputs, workflows, and data bundles quickly enough to hurt margins. S&P's defense is that even in the vulnerable segment, users are not buying raw data alone. They are buying curated outputs inside tools they already use. Management's Investor Day language centered on advancing essential intelligence, differentiated data, and workflows, with AI integrated into those tools rather than treated as an external threat. That matters because the company is trying to remain the interface, not get bypassed by it.
Why the cash cows likely hold for now
The line where the moat likely holds is simple: where trust, regulation, and embedded usage matter more than convenience.
That describes Ratings, Indices, and Energy. Those businesses are not just data vendors. They are infrastructure layers. Once benchmarks, rating frameworks, and commodity price references are embedded in pricing models, risk systems, and investment mandates, switching is harder than it is for a research assistant tool. That is why the moat rating and fair value remain intact despite AI anxiety around capital markets information services.
The financial backdrop also matters. In 2024, S&P Global produced $14.2 billion in revenue, up 14%, along with $5.58 billion of operating profit, up 39%, and $4.17 billion of net income, up 44%. Those are not the numbers of a business whose profit pool is being quickly disintermediated. If AI were only nibbling at the edge, the stock's panic would be out of step with the income statement.
The real test is adoption inside client workflows
Bears will counter that competitive pressure can arrive in waves: first in lighter analytics, then in client workflows, then in the reasons clients pay for bundled intelligence in the first place. That is a real risk, especially if AI vendors compress the value of standalone data products.
So the watchpoints are concrete:
- Does Market Intelligence growth decelerate materially before the broader portfolio absorbs it?
- Are clients consolidating around S&P's AI-enabled workflows, or moving toward cheaper point solutions?
- Do Ratings, Indices, and Energy keep delivering the stable profit base that makes the market's broad AI discount look too aggressive?
For now, the evidence still points to a segmented threat. AI is real in Market Intelligence. It does not look equally real in the cash cows.
What investors should track next
The next few quarters matter less for AI headlines than for proof of adoption inside S&P Global's actual workflows. That is the shift investors need to make: from fearing disruption in the abstract to tracking whether clients keep paying for trusted intelligence where decisions get made.
The next clean read on the thesis comes from how management talks about demand, not just technology. At Investor Day, S&P framed its strategy around advancing essential intelligence, with customers needing leading benchmarks, differentiated data, and workflows as markets evolve. That wording matters. If AI is improving the product, revenue should stay intact or even consolidate into S&P's tools. If the product is being bypassed, management's language will start to sound more like a feature update than a workflow defense.
What would weaken the thesis
The cleanest invalidation signals are not AI demos. They are sustained softness in Market Intelligence, client migration toward cheaper point tools, or more aggressive AI risk framing in disclosures. If those do not show up, the market's broad discount likely stays overstated.