Schlumberger Ties Nvidia Atop a New S&P 500 AI Ranking

Schlumberger, a century-old oilfield-services company, just tied Nvidia at the top of a new ranking of how S&P 500 firms are adopting artificial intelligence, with both posting a perfect 100. The AI-Driven Enterprise (AIDE) Institute graded every member of the index on how aggressively it puts AI to work, and the leaderboard looks far less like Silicon Valley than the technology’s headlines usually suggest.

The roster of perfect scorers makes the point. Nvidia and the social-media owner Meta share the maximum with that oilfield giant and the online retailer Amazon, and just below them sit two power utilities, a water-treatment giant and the country’s biggest grocer. The catch is what the number measures, a limit the institute’s own founder is quick to flag: how hard a company is leaning into AI, not whether the bet has paid off.

How the AIDE Index Scores Each Company

Behind the ranking is Paul Cheek, a senior lecturer at MIT Sloan School of Management and the institute’s chief executive. Rather than ask executives to grade their own AI fluency, his team reads the digital trail companies leave in public view. The method is a form of open-source intelligence (OSINT, drawing conclusions from freely available information), pulled from earnings-call transcripts, job postings, patent filings, LinkedIn profiles and corporate communications.

Each company is graded on four pillars, every one scored up to 100:

• Literacy – how much executives and staff demonstrably know about AI
• Advocacy – how often leaders talk the technology up, on earnings calls and beyond
• Orientation – how far the company is prioritizing AI in its strategy
• Implementation – how much AI has reached day-to-day operations

The headline company score that put four firms at the maximum is the average of just two pillars, orientation and implementation. The full AIDE Index of S&P 500 AI adoption, which folds in all four across 11 sectors, drives the separate sector leaderboards. Either way the raw material is enormous: by the institute’s count it read 518,779 source documents and roughly 272 million words, which it bills as the largest dataset of its kind.

Its selling point is the absence of self-reporting. Surveys capture what leaders wish were true, the argument runs, while public hiring and spending signals are harder to fake. That makes the index a sharp map of corporate intent, though it also means a top score rewards the companies pushing hardest on AI even when the payoff has yet to show up anywhere.

Energy and Utilities Crowd the Top of the Table

Run down the twenty companies with the highest company scores and the surprise is how little of it reads as technology. Nvidia is there, of course. Around it sit an oilfield contractor, a grocer, two power utilities, a materials company and two more energy names. Six of the top 20 slots went to energy and utility companies alone.

Here is where ten of the highest scorers land, sorted by each company’s S&P 500 sector.

The institute’s sector leaderboards, which use the full four-pillar index, spread the surprise wider still. Schlumberger tops energy and AES leads utilities, while Ecolab takes materials, Walmart consumer staples, Equinix real estate and Block financials. Microsoft, not the chipmaker, leads information technology on that broader measure. The throughline is that AI adoption signals now run hot in corners of the market that have nothing to do with building models.

Big Tech Still Sets the Ceiling

None of this means the chipmakers and platforms have been knocked off their perch. The perfect scores still cluster among the mega-caps that build and sell AI for a living, and the orientation pillar naturally rewards firms whose entire strategy is the technology. Nvidia, which became the world’s most valuable company on the back of AI accelerators, is the only information-technology name to post a flawless company score, and its drive to push that silicon into everything from data centers to its first Arm-based Windows laptop chip keeps its hiring and patent signals near the redline.

Amazon and Meta reaching the maximum fits the same logic, since both run AI through their own products at a scale few rivals can match, even if index bookkeeping files them under consumer discretionary and communication services rather than tech. The striking part is the company they now keep. Behind the familiar names sit drillers, utilities and a supermarket chain, close enough on the scoring to share the same tier.

Why an Oilfield Giant Out-Scored Most of Silicon Valley

So how does a company that drills for hydrocarbons end up level with the company that makes AI chips? The short version is that SLB has spent years turning artificial intelligence into a product it sells, pushing well past the pilot stage that traps most companies.

SLB Built AI for the Oil Patch

In 2024 the firm launched Lumi, a data and AI platform built for the energy industry, layering generative AI over the tangled subsurface and operational data that oil and gas runs on. SLB’s Lumi data and AI platform was pitched on domain depth, an AI tuned to a century of oilfield work rather than the open web.

AI is fundamentally altering the dynamics of our industry, but its transformational potential is hindered by the complexity of our industry’s data ecosystems.

That was Rakesh Jaggi, SLB’s president of digital and integration, on the day the platform launched. The wager is that whoever untangles energy data first gets to sell the intelligence built on top of it. It is also exactly the orientation-and-implementation signal the AIDE score rewards: real spending, real hiring and real patents, all pointed at AI.

Utilities Are Riding the Data-Center Demand Curve

The utilities are a subtler case. AES, NextEra Energy and Duke Energy all landed in the top 20, and part of that reflects genuine AI inside the business, from forecasting demand to balancing loads and flagging equipment failures on the grid.

But power companies are also among the loudest voices on AI, for a reason that has nothing to do with their own software: they sell the electricity that data centers drink. NextEra’s planned $66.8 billion takeover of Dominion Energy, among the largest utility deals in US history, was pitched almost entirely around AI-driven power demand. “Electricity demand is rising faster than it has in decades,” chief executive John Ketchum said when the deal was announced.

That demand story floods earnings calls and job listings with AI references, which is precisely what an outside-in index picks up. It is a real signal and a slightly misleading one at the same time. A utility can rank as an AI adoption leader partly because AI has become its biggest customer, not because it has rewired its own operations around the technology.

What a Perfect Score Leaves Out

Here is where the institute hits the brake itself. The index measures how hard companies are pushing on AI, and it says nothing about whether the pushing works; by design, it does not track financial returns. Independent research suggests that gap is wide, and the most-cited figures come from a separate MIT group, Project NANDA.

• 95% of enterprise generative-AI pilots delivered no measurable profit-and-loss (P&L) impact, Project NANDA found in its 2025 study.
• $30 billion to $40 billion in enterprise AI spending sat behind that result.
• Two of nine sectors studied, technology and media, showed real structural change; the other seven mostly experimented.
• Zero is the weight the AIDE company score gives financial performance, by design.

The Returns Question Stays Open

The contrast with the AIDE leaderboard is the whole point. An index built on hiring, patents and earnings-call talk lights up for any company spending heavily, whether or not that money ever lands in profit. The talk is at least getting more concrete: by one Morgan Stanley analysis of earnings transcripts, about a quarter of the S&P 500 cited a measurable AI benefit early this year, up from roughly one in eight a year earlier.

Even so, the institute does not dispute the distinction; it draws the line itself. Its index is a measure of effort and intent, an early read on which firms are reorganizing around what it calls the AI-driven enterprise model. Whether that seriousness converts into earnings is a question for later quarters and a different dataset.

The Bills Are Already Landing

And the spending is real even where the return is not. Companies that went all in on chatbots, coding assistants and autonomous agents are opening invoices several times larger than they signed up for, as the era of venture-subsidized AI pricing winds down, a squeeze we traced in the reckoning over soaring AI token bills.

The disconnect shows up at the high end of the AIDE table. Uber said it burned through its entire annual AI budget within four months, and Meta, a perfect scorer here, spent part of the year pressing staff to run up AI usage before reversing course once the bills arrived. High orientation, it turns out, can get expensive long before it gets productive.

How Boards Are Reading the Scorecard

For all those caveats, the index is built for one reader in particular: the board director who wants to know how the company stacks up without taking management’s word for it. That is the use case its creator, an MIT Sloan senior lecturer, keeps coming back to. “I don’t want it to be speculative,” Cheek said of the moment a board asks how it compares with its peers. “I want there to be some data that they can use to back up what they have to share.”

He also worries that the people reading the scorecard are underprepared. There is “significant room for improvement,” he says, in how well directors and executives grasp AI themselves, especially as it bears on managing risk and on the strategic investments that decide which companies create lasting value.

The scoreboard is open-source, and the institute plans to keep refreshing it, so next year’s edition will show whether today’s leaders turned orientation into output. If the energy and utility names crowding the top can start attaching real dollars to all that AI activity, the leaderboard will read like an early call. If they cannot, a score of 100 will end up marking the year a company talked the loudest about AI, in the same stretch that the spending got hardest to justify.