If you think of Microsoft’s moats as 1) massive switching costs 2) ecosystem bundling 3) network effects and inter-company compatibility 4) economies of scale, then AI certainly has the potential to both erode and strengthen all of these. Initial signs through my own work suggest that AI strengthens the moat especially regarding part 2, but also 3 and 4. Some believe AI will lower point 1), but I haven’t personally seen any significant signs of this yet. For now, I’d say we are clearly in the black.
AI services are certainly the main reason why Azure (and now later GCP) have grown faster than the market and gained share. To put it crudely, after 2022 Azure was what AWS was during the SaaS startup boom. Now the race has leveled out over the last 6 months based on my own empirical observations.
Microsoft is in a very good position regardless of the outcome. Practically the entire stack is covered. I don’t see this as such a significant factor. More important is how Microsoft can continue to capture a slice of the value it creates in the future, when the correlation between the license and the generated added value inevitably breaks as (agentic) AI becomes a part of, or even displaces, human-computer interaction.
– –
AI brings uncertainty to the continuity of a proven and lucrative business model. In my view, that is what the market is currently punishing with valuation compression. The door is opening for challengers, but I wouldn’t personally carry any particular concern for Microsoft regarding competitors taking slices of the market.
PS. why would the OpenAI deal be bad for MS? Okay, there have been disputes, but even those have stemmed from the fact that the deal has been so overwhelmingly good for Microsoft. I genuinely can’t think of almost anything bad from MS’s perspective, and I’ve written before that history will show if the deal was even better than the $75k deal for a certain piece of software called Quick and Dirty Operating System from 1981, which the old-timers know later as MS-DOS.
Don’t throw in the towel regarding quality. I have also always considered Microsoft to be super high-quality and regretted not picking it up during any dips or periods of low valuation. Well, now I bought some under $400. I don’t understand the threat of “SaaSgeddon” in the sense that Microsoft would just stand idly by while being overtaken from the left and right. If the massive investments of the coming years (future??) don’t yield as much as one would hope, they are unlikely to go completely to waste. And money keeps coming in anyway, and then we move forward. The ecosystem is extensive and the products are good. They know how to monetize and ensure customer retention. Azure is popular among enterprises, and cloud margins are solid. Companies use the products and are moving to the cloud more and more from their own servers. When buying Alphabet, the angle back then was that the value would rise WHEN Cloud becomes profitable (which happened); now we don’t have to wait for Azure’s profitability So now we’re just churning out cash flow and hoping for a good ROIC on the investments.
Perceptions have certainly changed immensely in recent years. This bumbling with Windows is embarrassing, but unless it shows up in the cash flows, weak sentiment might even provide a buying opportunity. For the rest of us, it provides a continuous stream of entertainment comparable to Netflix
Microsoft is supporting Anthropic in its legal battle against the Pentagon and is calling for at least a temporary injunction to prevent the immediate exclusion of the company’s AI technology from defense contracts.
According to Microsoft, a so-called “sudden stop” could disrupt the military’s existing systems. The dispute involves disagreements over how Anthropic’s models may be used for military and surveillance purposes.
Key Points
Microsoft threw its support behind Anthropic on Tuesday and advocated for a temporary restraining order that would block the Pentagon’s supply chain risk designation.
The move would “enable a more orderly transition and avoid disrupting the American military’s ongoing use of advanced AI,” Microsoft said in a filing.
Microsoft is negotiating with Chevron and Engine No. 1 for a large energy hub in Texas for data center needs, as growing AI demand increases investments and electricity needs.
Microsoft plans to invest approximately $10 billion in Japan between 2026 and 2029 to develop AI infrastructure and cybersecurity.
The company will train one million experts, collaborate with businesses and authorities, and aim to address the growing labor shortage in the AI sector.
The investment includes the training of 1 million engineers and developers by 2030, Microsoft said, which was unveiled during a visit to Tokyo by Vice Chair and President Brad Smith. In a statement, the company said the plan aligns with Prime Minister Sanae Takaichi’s goal to boost growth through advanced, strategic technologies while safeguarding national security.
The original message was in French, not English, which is why there’s that translated text there.
I haven’t really thought much about Copilot before, but this tweet makes some interesting points about it.
For example, among Microsoft’s 450 million 365 users, there’s enormous potential, as currently only 3.3% use Copilot. According to the tweet, the Copilot Cowork agent changes the game by automating work using enterprise data. The extremely large distribution power and additional sales potential of this are still, according to the tweeter, overlooked by the market.
Microsoft is renting a lot of additional server capacity in Narvik, Norway, to leverage Nvidia’s new chips.
Originally slated for OpenAI’s Project Stargate, the site shifted to Microsoft due to OpenAI’s cost-cutting pressures and failed negotiations.
Meanwhile, Google is expanding in London, while Microsoft is investing billions in infrastructure to meet the growing demand and capacity shortage for AI services.
Microsoft will rent 30,000 additional Nvidia Corp. Vera Rubin chips from neocloud provider Nscale at a campus inside the arctic circle in Narvik, Norway, Nscale said in a statement. This builds on a prior $6.2 billion commitment Microsoft made at the same site.
Microsoft surprised everyone by lowering Xbox Game Pass prices, even though, according to the article, games and subscription services are generally becoming more expensive.
Ultimate and PC Game Pass are getting cheaper, but on the other hand, new Call of Duty games will no longer be available on the service immediately upon release, but only after about a year.
The change is one of the first major decisions from new Microsoft Gaming CEO Asha Sharma, who took on the role after long-time CEO Phil Spencer, who was largely beloved by the gaming community, stepped down in February, and will likely earn goodwill among gamers.
Sharma already gained points with players in her first statement as CEO saying that she would not “chase short-term efficiency or flood our ecosystem with soulless AI slop.”
Microsoft and OpenAI have revised their agreement. Microsoft no longer has exclusive access to OpenAI’s AI models; instead, OpenAI can also sell them to others, such as Amazon.
In any case, Azure still remains the most important cloud service, but investors are nonetheless concerned about whether this weakens Microsoft’s position in the AI race.
Part of the trouble for Microsoft stems from its ability to serve AI customers. In its last quarter, the company said its Azure business revenue would have grown 40% if it had enough data center capacity to meet demand. Instead, Azure revenue grew 38%. This growth rate will be closely watched by investors on Wednesday.
I see these Microsoft and OpenAI contract arrangements somewhat like this restaurant owner analogy:
Old deal:
Hey, we’ll build restaurants for you if you pay us a portion of your revenue as rent, and you don’t rent from others anything except storage space or such from which you don’t sell food to the public. We’ll then buy food from you whenever we need it.
New deal:
Hey, we have so many other restaurant owners coming in that we won’t have time to build enough space for you. How about we make it so that you can rent restaurant space from others as well, but we take a cut of that revenue too, up to a certain cap. Additionally, we get to eat at your restaurants for free. At least you’re no longer just our property maintenance slaves.
Oh right, and we own 27% of you and we have the rights to your menu.
The company grew significantly and beat expectations quite broadly, especially in cloud services. Azure was the most important success in the report, which suggests that AI demand is genuinely supporting real business operations.
The enterprise customer backlog also looks very strong, which should provide good visibility into the future. However, concerns remain regarding the level of investment required for AI—a topic that has been discussed here as well—and the fact that near-perfect execution is already expected from Microsoft.
The company performed and continues to perform excellently, but a lot was and is priced into the stock.
According to the article below, based on the trial, the relationship between Microsoft and OpenAI appears to have been both a massive opportunity and a risk for Microsoft.
Satya Nadella feared early on that Microsoft would remain merely a “platform” for OpenAI, while OpenAI would rise above it in the tech hierarchy.
Microsoft benefited from AI infrastructure and Azure demand, but has still been seeking its own strong “AI path.”
Discovery in the high-profile case showed that Microsoft CEO Satya Nadella was worried about OpenAI supplanting his company in the tech hierarchy as far back as April 2022, seven months before the launch of ChatGPT, the event that kicked off the generative AI boom and turned Altman into a household name.
“I don’t want to be IBM and OpenAI to be Microsoft,” Nadella wrote in a email to executives that April, roughly three years after Microsoft wrote its first $1 billion check to OpenAI.
Nadella was referring to an earlier technology era, when Microsoft became more important than IBM, the dominant computer maker at the time. In 1980, IBM agreed to distribute Microsoft’s operating system on its computers, only to see the software maker eventually suck up the lion’s share of market value.
Anthropic was reported to be negotiating the use of Microsoft’s custom Maia AI chips to run its Claude models.
For Microsoft, this would be an important demonstration of the functionality of its own chips and would also reduce dependence on Nvidia. In turn, Anthropic would gain more computing capacity and options for running its models.
For Anthropic, renting Maia chips provides additional options to run its models and the opportunity to tailor future generations of the chip to its specific needs.
The Maia chips are designed to run existing models faster than Nvidia’s hardware, though they are not intended for training or developing new models, according to the report.
The article below explains that Microsoft is now developing more of its own AI models so that it doesn’t have to rely as much on OpenAI. The idea is that Microsoft can run these models on its own servers and then offer them to developers at a lower cost.
The new model helps generate programming code from “text instructions,” meaning Microsoft wants a bigger slice of the AI market, among other things.
After refining its models for the needs of consulting firm McKinsey, Microsoft was able to outperform OpenAI’s GPT 5-5, with 10 times better cost efficiency, said Mustafa Suleyman, CEO of Microsoft AI.
The coding model is “inference ultra-efficient,” Daigle wrote, and is available in the GitHub Copilot AI coding service and the Visual Studio Code text editor.
Also on Tuesday, Microsoft is revealing updated cloud-based models for speech recognition, synthetic voice generation and image generation, as well as small Aion models that can run on Windows PCs.
Key Points
At its Build developer conference in San Francisco, Microsoft announced MAI-Code-1-Flash, its inaugural model in the AI coding space.
Microsoft is trying to establish a presence with proprietary models to compete with OpenAI, Anthropic and Google.
Microsoft’s primary role in the AI boom to date has been as a provider of cloud infrastructure and services and as an investor.
Xbox is facing major layoffs and cost-cutting measures as profitability has clearly weakened. Console sales are slumping, Game Pass growth has stalled, and hit games haven’t been released at a consistent enough pace.
Microsoft now intends to revamp Xbox’s strategy and keep only some of its games exclusive to its own platforms.
So now we’re just churning out cash flow and hoping for a good ROIC on the investments.
