Microsoft Reassesses AI Strategy Amid Copilot Challenges and Model Competition

Editor’s Note: Microsoft was among the first giants to bet correctly on the generative AI wave by investing in OpenAI. With its investment in OpenAI and exclusive cloud partnership, Microsoft was once seen as the most certain winner of the AI era: Azure reaped the benefits of model advancements, while Office, Bing, GitHub, and its enterprise software lines were fully integrated with Copilot. Nadella, much like he led Microsoft’s shift to cloud computing, was expected to once again orchestrate a platform-level transition.

But two years later, Microsoft’s advantages began to grow more complex. OpenAI was no longer just a technology supplier to Microsoft—it also became a direct competitor for enterprise customers; models like Claude and Gemini rapidly caught up, eroding the lead once granted by GPT’s exclusivity; and the emergence of AI agents further disrupted Microsoft’s long-relied-upon SaaS business model. Stock price corrections, lower-than-expected paid adoption of Copilot, and GitHub Copilot being overtaken by Cursor and Claude Code have all forced Microsoft to reassess its AI strategy.

What’s most noteworthy about this article isn’t whether Microsoft can still catch up to OpenAI, Anthropic, or Google in model capabilities, but rather that Microsoft is attempting to redefine its position: it is no longer betting everything on a single model, but instead shifting toward a “model-agnostic” enterprise AI platform strategy. In other words, Microsoft aims to become the foundational layer connecting models, data, security, workflows, cloud computing, and enterprise software. The models themselves may come from OpenAI, Anthropic, or even Microsoft’s own Superintelligence team in the future—but what truly remains within Microsoft’s ecosystem are enterprise customers’ work platforms, data assets, development environments, and security frameworks.

This is also the context behind Nadella personally engaging in Copilot’s product development. For Microsoft, the AI competition is no longer just a model race between labs—it’s a systemic battle over organizational speed, product form, customer relationships, and capital expenditure. Claude Code and Claude Cowork demonstrate that AI agents could reshape software development and workplace workflows; open-source projects like OpenClaw show that a “always-on” AI assistant is moving from concept to reality. Microsoft’s task is to package these more aggressive AI-native experiences into security, compliance, and governance frameworks that enterprise customers can accept.

However, the cost of this path is not low. To keep pace with cutting-edge models and support agent-based products, Microsoft is pushing the AI competition toward “gigawatt-scale” infrastructure investment: more data centers, larger chip clusters, and higher capital expenditures. By 2026, Microsoft expects its capital expenditures to reach approximately $190 billion. In other words, in the AI era, Microsoft must move like a startup—rapidly experimenting and iterating—while also sustaining the heavy asset investments of a cloud computing giant.

The real challenge Microsoft faces is not whether it can still be the sole winner in the AI era, but whether it can continue to maintain its core entry point in enterprise software amid the rapid commoditization of models and the ongoing disruption of software business models by agents. For Nadella, this may not be a routine product adjustment, but rather Microsoft’s second entrepreneurial journey during its transition to an AI platform.

The following is the original text:

Mid-January 2026, Redmond, Washington. The weather was cold and gray—a morning perfect for hitting the snooze button. But inside Building 92 of Microsoft’s sprawling campus, a team of engineers had already arrived early.

They are fighting a tough battle and are already behind.

The team is developing a new AI product—it’s more like a personal assistant that can help users book flights, reply to emails, and even find reliable local plumbers. The team members are well aware that other tech companies are also developing similar products. Just then, Microsoft CEO Satya Nadella arrived on the scene. He wanted to show them something.

Nadella opened his laptop and launched an application—a system for orchestrating and controlling multiple AI agents, which he called the "Chain of Debate." As he demonstrated, he explained it to the engineers. Team members exchanged knowing glances, like seasoned basketball players suddenly realizing a new player is actually really good.

This app wasn’t made for Nadella by someone else; he built it himself using AI tools through “vibe coding.”

“This set the tone for how hard the entire team would need to push forward,” recalled Jacob Andreou, Microsoft’s executive vice president responsible for Copilot design. At the time, Nadella was in the same room with everyone, almost standing behind the engineers and even opening his own computer to join in.

The team was energized by the CEO’s excitement over building a new product from scratch. By the end of February, this sprint concluded with Microsoft launching Copilot Tasks—an AI-powered personal assistant tool capable of using a computer. Nadella’s earlier prototype served as a reference model for a feature called “Model Council” and other components within Copilot.

But Nadella’s frequent deep involvement with AI product teams—even personally building prototypes—speaks volumes about Microsoft’s current situation. After all, this is a $3 trillion tech giant, not a grassroots startup where the CEO regularly joins developers on sprint cycles to write code.

Nadella’s concerns about Microsoft’s AI strategy have been clear enough. In October last year, he announced that he would step back from some of his business responsibilities to focus more on AI research, product innovation, and AI data center development.

This concern is not unfounded. Microsoft’s stock previously went through a difficult period; after reaching a historical high in October last year, its share price fell by approximately 34% over the following five months. Meanwhile, Microsoft’s cloud computing platform Azure and AI-related revenue more than doubled over the past year.

Microsoft has also become one of the classic victims of the so-called “SaaSpocalypse”—a mass sell-off of SaaS stocks triggered by the emergence of AI programming agents. Many investors now believe that such products mean enterprises will no longer purchase AI tools from SaaS providers like Microsoft, or even off-the-shelf software in the future.

Between October 28, 2025, and March 27, 2026, Microsoft’s stock price declined by 34%. Sales of the Microsoft 365 Copilot enterprise product have also fallen short of the company’s expectations. Among the 450 million users of the Microsoft 365 suite, fewer than 4.5% are currently paying for Copilot features. Meanwhile, consumer usage of the Copilot chatbot lags significantly behind ChatGPT, Gemini, and Claude. Even GitHub Copilot, once the leading AI programming assistant, has been surpassed by AI startups Cursor and Anthropic’s Claude Code.

Two years ago, Microsoft appeared to be one of the earliest winners of the AI era. Thanks to Nadella’s forward-looking bet on OpenAI, Microsoft gained exclusive access to the startup’s rapidly growing AI models and integrated them into its own product ecosystem. If enterprises wanted to use OpenAI’s technology, Microsoft Azure was the only cloud provider available. Microsoft even once believed that OpenAI offered its most promising opportunity in years to challenge Google Search.

At that time, Nadella had been leading Microsoft for a full decade. He previously guided Microsoft’s transition from desktop software to cloud computing, and now appears poised to replicate this success in the AI era.

But AI is changing too rapidly. Two years is enough to constitute a long cycle. The next chapter is how Microsoft lost its early AI lead and how it is now trying to regain the initiative.

Microsoft initially rose to the forefront of the AI race thanks to its partnership with OpenAI; but part of what has put it on the defensive is precisely this partnership.

Microsoft recognized the young San Francisco company early on and made its first investment of $1 billion in 2019, later committing a total of $13 billion to OpenAI. Leveraging OpenAI’s technology, Microsoft has launched a series of AI products branded as Copilot across its consumer and enterprise software offerings.

But after the release of ChatGPT at the end of 2022, OpenAI’s explosive growth and rapidly expanding ambitions soon strained the relationship between the two companies. They clashed on multiple issues: in computing resources, OpenAI consistently demanded more; in intellectual property, Microsoft felt OpenAI had not fulfilled its contractual obligations promptly by sharing technological innovations; in customer relations, OpenAI began directly marketing its AI models to the same enterprise clients to whom Microsoft was also selling Copilot; and when OpenAI sought restructuring, the two parties disagreed over how much equity Microsoft should receive in the new for-profit entity.

Nadella knew that betting Microsoft’s AI strategy on an unproven startup carried inherent risk. In November 2023, that risk became starkly apparent: OpenAI’s nonprofit board, which controlled its for-profit operations, fired CEO Sam Altman for “failing to remain consistently candid,” notifying Nadella only minutes before publicly announcing the decision.

Nadella had to quickly reassure investors, emphasizing that Microsoft still retained access to OpenAI’s technology; meanwhile, he collaborated with Altman to pressure the board into reversing its decision. Nadella announced that Microsoft was prepared to hire Altman, along with any OpenAI employees willing to join him. The prospect of a mass exodus of staff ultimately forced the board to back down and reinstate Altman.

Inside OpenAI, the five-day crisis later became known as "the blip." But according to people familiar with Nadella’s thinking, the event deeply shook him—he needed to find a hedge for Microsoft’s AI bet.

When Nadella joined a sprint with the company’s AI engineering team, it set the tone for how hard the entire team would push forward.

—Jacob Andreou, Executive Vice President of Microsoft Copilot

Microsoft's Plan B is Mustafa Suleyman.

Suleiman was a co-founder of Google DeepMind and later left to found his own AI startup, Inflection. In March 2024, Microsoft hired Suleiman and Inflection’s technology team in a $650 million deal, gaining licensing rights to their technology. Subsequently, Suleiman was appointed CEO of Microsoft’s new AI division, known as MAI, which has two key responsibilities: developing cutting-edge models internally at Microsoft as a hedge against OpenAI, and expanding the user base of Microsoft Copilot.

However, progress on this step was not smooth. Microsoft’s partnership agreement with OpenAI prohibited Microsoft from training models beyond a certain scale. Sulaiman told Fortune: “At the time, we were essentially limited to training Microsoft’s own native models, and only up to the scale of SLMs—small language models.”

The first publicly tested general-purpose language model from MAI, named MAI-1 Preview, was released in August 2025 but ranked relatively low on various performance leaderboards and was ultimately never widely released.

MAI also failed to turn Copilot into a consumer hit. According to media reports, a year after Sulaiman took over, Copilot’s usage plateaued at around 20 million weekly active users, while ChatGPT’s user base surged, eventually reaching 900 million. In 2025, Microsoft attempted a major upgrade to Copilot, aiming to make it more like a task-executing personal assistant, but the update failed to reignite growth. Meanwhile, the new AI-powered version of Bing search has barely made a dent in Google’s market share.

Meanwhile, Plan A also began to encounter difficulties.

In 2023, OpenAI’s GPT models led the industry by a wide margin. But by early 2025, Anthropic’s Claude had frequently topped AI rankings, and many enterprises began preferring it for complex tasks. Google’s Gemini also became increasingly competitive in visual tasks. Meanwhile, Microsoft’s Copilot products remained entirely powered by GPT. The engine that once underpinned Microsoft’s AI strategy had begun to feel like a heavy burden.

Microsoft Business CEO Judson Althoff acknowledged that the company made several mistakes. First, naming both consumer and enterprise products "Copilot" was inherently confusing. Althoff, who holds a private pilot’s license, joked, “Worse than having no co-pilot is having more than one co-pilot.”

Microsoft also encouraged sales representatives to promote both the freemium and premium versions of M365 Copilot for enterprise, but only the premium version truly delivers value to enterprise customers. "We got this wrong," he said.

Microsoft is also striving to keep pace with the evolution of AI technology. A key turning point occurred in 2025, when Anthropic released Claude Code. Developers only needed to describe what they wanted, and it could autonomously write complete programs. This was no longer a "co-pilot," but "full autonomy." Within just six months, it transformed the way software is developed.

Later in January, Anthropic launched Claude Cowork, an agent capable of using software such as Microsoft productivity tools including Excel and PowerPoint to autonomously complete tasks.

Claude Cowork poses a serious challenge to M365 Copilot and the AI agents Microsoft has been actively promoting to its customers. In fact, it threatens not just Microsoft, but most business software. This realization triggered a sell-off in software stocks, known as the "SaaSpocalypse." Ultimately, the technology sector lost over $2 trillion in market value, including a single-day loss of $357 billion for Microsoft.

By autumn 2025, Nadella realized that Microsoft had to reboot its AI strategy. Since then, the company’s actions have reflected a difficult balance: on one hand, it must innovate rapidly like an AI startup; on the other, it must continue to serve investors and enterprise customers with the same reliability as the traditional Microsoft.

Nadella delegated many business and day-to-day operational responsibilities to Microsoft senior executive Althoff, allowing himself to focus on AI product development. Althoff said he is responsible for "Horizon Zero" and "Horizon One," while Nadella oversees "Horizon Two" and "Horizon Three." Meanwhile, Nadella began breaking down internal barriers to make Microsoft faster, flatter, and more agile.

In March this year, Nadella merged the consumer and enterprise Copilot teams. Sulaiman no longer oversees consumer AI products but now leads a renamed model development initiative: the Superintelligence team. Sulaiman said the name reflects the team’s ambition and helps attract top researchers.

Jacob Andreou joined Microsoft in 2025, following prior roles at Snap and the venture capital firm Greylock. Today, he leads the Copilot experience for both consumers and enterprises and reports directly to Nadella. Joining Suleman and Andreou on the Copilot leadership team are three other senior Microsoft executives: Charles Lamanna, who oversees Copilot, AI agents, and platforms; Ryan Roslansky, who leads Microsoft Office and LinkedIn; and Perry Clarke, Chief Technology Officer for Applications.

Lamanna said, “We want it to be a backend, a brain, driving both consumer and workplace experiences.” Nadella himself attends the weekly stand-ups of the Copilot leadership team and participates in a dedicated Teams channel focused on the ongoing development of Copilot.

Microsoft faces a delicate balance: it must innovate quickly enough to keep up with AI competitors like Anthropic and Google, while continuing to be seen as a reliable partner by large enterprise customers.

Andreou noted that the two new products demonstrate that the unified Copilot team is operating as Nadella envisioned: one is Copilot Tasks, a consumer-facing product that Nadella personally helped prototype in January; the other is Copilot Cowork, designed for enterprise customers.

He said: "Both products have essentially achieved cutting-edge experiences—one targeted at consumers and the other at enterprise users. And both were rapidly developed by our team by integrating resources within just a few weeks."

Microsoft has also agreed to OpenAI’s long-awaited restructuring plan, with significantly reduced restrictions. The software giant has acquired a 27% equity stake in OpenAI. If OpenAI goes public, as widely expected, this could provide Microsoft with substantial upside potential. However, the exclusivity arrangements in the original agreement have been abandoned: OpenAI can now partner with other cloud providers, and Microsoft can also use AI models from other companies.

Suleiman said the new agreement finally allows Microsoft to build larger, more capable frontier AI models and eventually achieve self-sufficiency. However, he added that Microsoft still needs two to three years to catch up with the top AI labs.

The renewed partnership also enabled Microsoft to embrace Anthropic, a major competitor of OpenAI. In November last year, Microsoft committed to investing up to $5 billion in Anthropic and began offering its models on Azure. The ability to power Copilot with Claude has been well received by enterprise customers and has helped Microsoft develop Copilot Cowork.

“We must acknowledge that OpenAI and Anthropic are helping us run faster.” — Judson Althoff, CEO of Microsoft Commercial

But Microsoft is not simply replacing its reliance on one loss-making AI startup with reliance on another. Behind its investment in Anthropic is Microsoft’s different assessment of industry trends: AI models will become increasingly commoditized. At least in the enterprise market, true value will no longer reside solely in the AI’s “brain,” but will shift toward the tools, data, security, cloud computing, and workflow systems that surround it.

This is precisely where Microsoft believes it can win.

It already possesses many key assets: software tools, security systems, data warehouses, and cloud computing capabilities. Microsoft has also developed a suite of IQ-branded products that help businesses create customized workflows, consolidate their data, and build, deploy, and monitor agents that run these workflows using any AI model from any vendor.

Althoff said: "We don't believe companies will replace their information work platforms, development environments, and security environments every time a new model is released."

This strategic shift has also brought about new business models.

In the past, Microsoft typically charged users on a per-user basis, such as $30 per user per month for Copilot. Customers favored this model because it made budgeting easier. However, if the AI agents in these products use models that Microsoft does not own, Microsoft must pay the AI providers for the corresponding token consumption.

Therefore, Microsoft has begun shifting to a hybrid pricing model: the base tier continues to be charged per user license and includes a limited token quota, while usage beyond that is charged per token. This approach is designed to prevent the "model-agnostic" strategy from eroding profit margins.

In consideration of cost control, Microsoft has also begun streamlining its workforce. In April this year, Microsoft announced its first-ever voluntary separation program, primarily targeting employees with the longest tenure. The company stated that approximately 7% of its U.S. employees—about 8,750 people—are eligible for the program, with an expected cost of $900 million.

There are signs that Microsoft’s adjusted corporate strategy is working. As of the end of March, Azure revenue grew 40% year-over-year, and Microsoft’s overall AI business reached an annualized sales run rate of $37 billion, up 123% year-over-year. Currently, 20 million M365 users are paying for Copilot, with a quarter of them added in the first four months of 2026. Althoff noted that adoption is accelerating.

UBS analyst Karl Keirstead said that an increasing number of Microsoft customers have told him they are beginning to see the value of Copilot, but overall user adoption remains disappointing. He said, "I don't think they've reached the level of penetration that Wall Street would find satisfactory."

Microsoft’s “model-agnostic” strategy may also have a vulnerability: what if those highly publicized AI startups begin building enterprise tools and integration systems similar to Microsoft’s?

This is no longer hypothetical. In February of this year, OpenAI launched the Frontier platform for enterprises, offering many of the capabilities Microsoft is building into its new tools. Anthropic is also moving in this direction, introducing its Claude Managed Agents service.

Microsoft's argument is that decades of enterprise customer relationships, its reputation for reliability and security, and deep integration with customers' existing software systems will give it an advantage. Althoff said he welcomes the competition. “It must be acknowledged that OpenAI and Anthropic are helping us run faster,” he said.

However, some have questioned whether a company the size of Microsoft can truly match the agility of AI-native startups. UBS’s Keirstead said, “Microsoft, and frankly all software companies, are facing a situation they haven’t encountered in over a decade: radically innovative new competitors. It may be unreasonable to expect large established players like Microsoft to pivot as quickly as OpenAI and Anthropic.”

Bank of America analyst Tal Liani sides with the Nadella camp. He believes AI companies are unlikely to build the full suite of products that Microsoft offers, meaning Microsoft doesn’t need to win the AI race—it just needs to avoid losing it.

He said: "It doesn't have to be the best—just good enough, and by bundling it to deliver high value, that's really Microsoft's value proposition."

However, even just "breaking even" comes with a high cost.

Like other hyperscale cloud providers, Microsoft is investing heavily in data centers and custom chips. In fiscal year 2025, Microsoft’s capital expenditures reached $88.2 billion, roughly in line with peers such as Google Cloud and Amazon AWS. However, in hindsight, this level of investment was still too conservative. Surging demand has left Microsoft with insufficient computing capacity and unable to recognize signed AI revenue as actual revenue at the expected pace.

"I thought we would catch up," CFO Amy Hood admitted on the earnings call last October, "but we did not."

Now, Microsoft is doubling down further. The company expects its capital expenditures in 2026 to reach approximately $190 billion, more than triple its 2024 spending. Wall Street, once wary of such spending levels, now appears willing to tolerate these massive investments. But if investor sentiment turns, Microsoft could be more vulnerable to risk than ever before.

In November 2025, an independent developer named Peter Steinberger released OpenClaw—a free, open-source system that transforms any AI model into a long-term, autonomous, always-on agent capable of developing software, acting as a virtual administrative assistant, or even managing inventory for an online store.

OpenClaw is highly popular among developers and early AI adopters. It is reported that Nadella is also one of them.

However, while OpenClaw is popular, it has a significant drawback: to function effectively, it requires access to your system, data, payment information, and passwords, making it extremely risky. Additionally, it consumes tokens at an astonishing rate.

In March this year, Nadella said at a tech conference in San Francisco: "I cannot launch OpenClaw at Microsoft; I don't have the authority to do so, because it would be seen as Microsoft releasing a virus. But at the same time, it is indeed an incredible innovation."

Nadella has asked the unified Copilot team to build Microsoft’s version of OpenAI: one that retains the fun and ease-of-use of consumer-grade products while incorporating the security and governance capabilities enterprises demand. Andreou sees this as a test for the new organization: “This is what we call victory here.”

Lamanna believes this could become the catalyst for Copilot's growth. He said, "The hardest question has always been: How do you help people change the way they work?"

If a continuously running AI assistant is truly feasible, it would make this shift easier to achieve. It also means that the fundamental unit of AI will shift from “models” to “always-on agents.” This is precisely a paradigm shift that will test whether Microsoft’s so-called “connected organization” strategy can still hold as its core form evolves. Lamanna said that an enterprise-grade Microsoft version of OpenClaw is already within reach.

During the week of March 30, Sulaiman gathered the new Superintelligence team in Miami for a three-day offsite meeting. The team, consisting of approximately 500 members from around the world, aimed to develop a roadmap for achieving “gigawatt-scale” AI training runs. This scale of training would enable Microsoft to directly compete with OpenAI, Anthropic, Google DeepMind, Meta, and xAI.

Suleiman stated that it is crucial for Microsoft to achieve self-sufficiency before 2030, as it will lose access to OpenAI's technology in 2032.

The entire team gathered in a large banquet hall to listen to Sulaiman and Nadella deliver keynote speeches and participate in an "Ask Me Anything" Q&A session. According to Sulaiman, Nadella described this moment as Microsoft’s effort to “reinvent the company” in response to the shift toward AI platforms.

This is a meaningful statement.

After the keynote speech, the conference split into different workflows. Teams gathered around the 40 whiteboards arranged around the banquet hall to brainstorm and plan the eight-week sprint ahead. Nadella did not leave; he stayed behind.

Over the next three hours, he moved between tables, speaking with researchers, offering suggestions, and sharing ideas.

If this is truly a "reinvention," then Nadella is acting like a startup CEO. He takes no advantage for granted. He knows Microsoft could lose everything—and still has everything to gain.