TheChek News Flow

A grimace is worth a thousand words.

The facial expression in question belongs to Microsoft CEO Satya Nadella earlier this week in an Oakland, Calif., courthouse as he took the stand in Elon Musk’s lawsuit against OpenAI.

Nadella’s discomfort is understandable. Here we have the usually sweater-clad executive in a suit and tie, testifying under oath about his private correspondences, in a multitrillion-dollar trial where he’s locking horns with the richest man in the world.

No doubt Nadella would rather have been at the dentist. And yet, fighting this fight is very much part of his remit, particularly since as of last October, Nadella turned over the keys of the commercial side of Microsoft (which accounts for more than 70% of its revenue) to Judson Althoff, putting him in charge of the most important part of the company.

“I am focused on horizon zero and horizon one,” Althoff tells me in high Microsoft-speak. “Satya tends to focus on horizon two and horizon three.” Translation: Judson’s running the current business. Satya’s going to figure this artificial-intelligence thing out.

The two endeavors are inextricably linked, as Microsoft hopes that business customers increasingly will adopt its Microsoft 365 Copilot AI product and other AI offerings. In other words, selling AI is very much a here-and-now matter for Microsoft.

Even though Microsoft says that Nadella will “be laser-focused on our highest-ambition technical work,” the CEO must also wrestle with massive strategic issues, like Microsoft’s relationship with OpenAI and its ChatGPT large-language model, the “brains” of its AI endeavors. Microsoft has invested nearly $12 billion in OpenAI over the past seven years, and now has a 27% stake in that company worth a staggering $230 billion.

Musk’s lawsuit is an existential threat to that relationship as he seeks to wind OpenAI back to its 2015 nonprofit roots—as well as undo its special relationship with Microsoft. (Microsoft also has a partnership with OpenAI rival Anthropic.)

And yet there may be even more to Nadella’s pained look than just battling Musk—that being Microsoft’s stock performance, which over the past year has declined by 9.6% versus gains of 138.1% for Alphabet and 27.3% for the S&P 500 index.

Of course, Microsoft shares over Nadella’s tenure, which began on Feb. 4, 2014, have been like a SpaceX rocket, up 1,025% versus 322% for the market. Nadella’s predecessor as CEO, Steve Ballmer, once told me, “I love Satya. I haven’t sold my stock.” (Ballmer reportedly owns some 300 million shares, now worth $130 billion. They were valued at just under $12 billion when Nadella took over as CEO.)

Microsoft’s recent quarterly numbers were up briskly almost across the board. The company reported adjusted earnings of $4.27 a share on revenue of $82.9 billion for its fiscal third quarter, while analysts expected $4.05 a share on revenue of $81.4 billion. Its Azure cloud business grew a stellar 40%.

On Friday, the ever-voluble hedge fund manager Bill Ackman disclosed in an 887-word tweet that he had recently acquired a position in Microsoft as “a core holding.”

Still, AI fears are making the Beast of Redmond, as Microsoft is sometimes called, look a little less fearsome. The Financial Times reported that Chris Hohn, the highly regarded manager of TCI Fund Management, sold most of its $8 billion stake in Microsoft, which it had held for the past 10 years. TCI, the most profitable hedge fund in the world last year with a sterling long-term record, noted in its letter to shareholders: “We reduced our investment in Microsoft because the rapid progress in AI introduces uncertainty over Microsoft’s competitive position in the future.” (It will be great fun to see who’s right here, Hohn or Ackman.)

Another risk to Microsoft posed by AI is prodigious capital spending on chips to run AI, data centers, and electric power generation. Capex has climbed from $24 billion in fiscal year 2021 to $88 billion in fiscal year 2025. The $190 billion projected this year is based on MSFT’s 2026 calendar year.

That multi-hundred-billion-dollar spend, plus Hohn’s concerns about AI making Microsoft’s legacy business obsolete and questions about AI adoption by Microsoft’s customers, are making Wall Street nervous and raising questions that Althoff, in his increasingly public-facing role, wants to address. (Note that Althoff, 53, may one day be in line to become CEO of Microsoft, though that would be years off, as Nadella, 58, isn’t likely to be going anywhere anytime soon.)

So, what does Althoff think about those AI worries?

“They don’t concern me, because I think the market is still trying to figure out AI,” he says. “When customers or analysts start thinking, ‘Gosh, AI is gonna eat software or SaaS,’ it comes from the place of seeing parlor tricks. It’s very easy to ask a model to create a document or spreadsheet. It is materially more difficult to say, ‘Take this presentation for a meeting with my board, update this data on our financial performance, and give me the six-trailing quarter view of a certain business and have it be in context, accurate, and pulling from secure sources.’ ”

As for the capital being spent, Althoff is equally sanguine. “Yes, it’s expensive,” he says. “I don’t get concerned. We’re building for our first-party AI solutions and infrastructure for Anthropic. We’re building for a future where the combination of human ambition and AI will come together to run the world’s business processes.”

Microsoft says it has 20 million paying seats for its premium AI Copilot product out of a total of some 450 million business seats. Tigress Financial Partners, which has a Buy rating on the stock and a price target of $680 (the stock currently trades at $407) notes that “adoption is scaling rapidly, with paid Copilot seats growing triple digits year over year to tens of millions of users.”

But what are customers actually doing with Microsoft’s AI products? Raj Sharma, global managing partner for Growth & Innovation at EY, has rolled out an agentic AI platform, called EY Canvas, to facilitate the company’s core global audit business, based on Microsoft AI products.

“If I looked at all the [technology] stack and the capital spend that is required, only three companies stood out to me for an enterprise of our size: Microsoft, Google, and AWS,” says Sharma.

“We picked Microsoft because of our longstanding relationship, and they will give us access to any models, whether it is OpenAI or Anthropic. Also, there was a level of trust with them that they will co-innovate with us.”

Rob Goldstein, COO of BlackRock, says his company worked with Microsoft and its AI tools to translate BlackRock’s websites into local languages around the world. “These are highly regulated websites that need to have all sorts of information about the company and the products we provide,” he says. “We feel very good about our relationship with [Microsoft],” but he adds, “this will not be a one-vendor game.”

All hasn’t been smooth sailing in Microsoft’s AI efforts. The Wall Street Journal reported that customers have been confused by various Microsoft AI brands and products, and that some have preferred using Google’s Gemini and other options. The company recently changed up leadership in its AI teams, presumably to address these issues.

To longstanding Microsoft watchers, as my colleague Adam Levine noted in a recent Barron’s cover story, the choppy start may be chapter one of a familiar narrative: Microsoft typically isn’t first, but the company’s vast installed base means it can switch into Beast Mode and win in the end.

That last point almost makes Althoff cringe. “We take nothing for granted,” he says. “While I am proud of our base and grateful to our customers for trusting Microsoft, I want to make sure that we continue to build the best AI capability. I always say to my folks, ‘Neither winning or losing are statements of perpetuity.’ ”

The bottom line is that companies want to know if shelling out big bucks today for AI tools, from Microsoft or its competitors, is truly worth it. Right now, the jury may still be out. In a new white paper, “LLMs Corrupt Your Documents When You Delegate,” researchers concluded that for longer tasks, “current LLMs are unreliable delegates: They introduce sparse but severe errors that silently corrupt documents, compounding over long interaction.”

The three authors of the paper are from Microsoft Research.

For investors with insomnia, Nasdaq CEO Adena Friedman has just the tonic, and it’s right around the corner: 23-hour-a-day trading. (Still only five days a week, though.)

I asked Friedman why this was necessary.

“Technology has really continued to advance to allow for investors from all over the world to be able to trade in U.S. equities anytime during the day,” she says. “Trading 23-5 is already happening. It’s just happening in the dark. Our systems today open at 4 a.m., and we allow trading to start then, and they close at 8 p.m. So, we are already trading outside of U.S. market hours. What we’re doing is basically expanding that.”

Traditional U.S. stock exchanges are open only from 9:30 a.m. to 4 p.m. Eastern Time, but extended trading has been growing through various means. Some institutions and individuals trade in lower-liquidity after-hour sessions. Some broker-dealers offer extended trading for certain stocks, and there are also derivatives that can mimic after-hours trading. Crypto and currencies trade 24-7.

Friedman notes that Nasdaq is partnering to facilitate longer hours. “We’ve worked with the industry to bring the consolidated tape and have that launch at the same time, which is Dec. 6,” she says. “You will have central transparency of all of the trading and quoting activity that’s occurring in the lit venues and dark pools around the world.” The exchange plans to add circuit breakers to halt trading if prices go up or down too much, “so we have more trading guardrails,” she adds.

Friedman notes that the Depository Trust & Clearing Corp. is extending its hours so there will be real-time clearing.

“There are millions of investors who want to access the U.S. economy and trade amazing companies, and this will give them a good infrastructure to be able to do that,” Friedman says. “We feel that it will increase overall demand and increase the pie. We’re working with a lot of clients who are looking at how to expand their infrastructure to support 23-5 trading, and we can provide technology to support them.”