Microsoft is looking a bit vulnerable. The company once claimed an early lead in AI through its OpenAI partnership. Now, it’s finding it hard to absorb pressure building across three fronts—Copilot, a fraying relationship with OpenAI, and questions around AI leadership under Mustafa Suleyman.
In what appears to be a reset moment, Microsoft has now shaken up Copilot leadership, unifying its enterprise and consumer segments under Jacob Andreou. He joined Microsoft AI in January 2025 as Corporate Vice President for Product and Growth, after serving as SVP of Product and Growth at Snap for eight years.
He is joined by Ryan Roslansky, Perry Clarke, and Charles Lamanna, promoted to head Microsoft 365 apps and the core Copilot platform.
Meanwhile, Suleyman, who co-founded DeepMind, will now pour his energy into building next-generation frontier models. “The next phase of this plan is to restructure our organisation to enable me to focus all my energy on our Superintelligence efforts and be able to deliver world-class models for Microsoft over the next five years,” he said on the appointments.
In 2024, Suleyman was appointed CEO of Microsoft AI, taking charge of consumer products such as Copilot, Bing, and Edge. However, under him, Microsoft struggled to keep pace with its rival, Google.
Microsoft hasn’t built any frontier models that could match AI giants. However, it has developed a few in-house AI models, including the Phi family, and is working on larger models such as MAI-1. Recently, it launched its image generator model MAI-Image-2.
Meanwhile, Copilot usage numbers remain relatively modest, especially compared to consumer AI chatbots. As of February 2026, the Copilot app had about 6 million daily active users, according to Sensor Tower analytics, whereas ChatGPT logged roughly 440 million and Google’s Gemini around 82 million.
To turn the tide, in November 2025, the company formed the MAI Superintelligence Team under Suleyman.
Microsoft should fire Mustafa Suleyman to be able to build models. Come. On. https://t.co/zHsjGLPjlM
— Ksenia_TuringPost (@TheTuringPost) March 19, 2026
Copilot Struggles Lore
On the enterprise side, only about 3% of Microsoft 365 commercial users have signed up for the paid Copilot add-on, or about 15 million. While that figure has grown 160% year-on-year, this is a small share, considering Microsoft 365 has around 450 million paid seats globally.
Copilot is still finding its footing.
“From a user’s perspective, the experience today feels fragmented,” Jacob Joseph, VP, Data Science, CleverTap, tells AIM.
“Copilot behaves differently depending on which Windows app you're inside of. It feels more like a disparate layer spread across products than a system that understands workflows end-to-end. Enterprises don’t adopt AI because it’s impressive in isolation; they adopt it when it’s predictable, reliable, and embedded into existing processes without forcing behavioural change,” he explains.
Businesses are still trying out Microsoft 365 Copilot, but the tech giant is releasing more products. In March, the company introduced a new premium enterprise plan, Microsoft 365 E7, priced at $99 per user per month, which is about a 65% increase, with Copilot bundled into the top tier.
“What Microsoft needs to get right over the next 12–18 months is less about adding more AI and more about tightening the product. Copilot has to evolve from a set of assistive features into something that feels like a dependable co-worker—consistent across surfaces, deeply integrated with enterprise data, and opinionated about specific, repeatable use cases,” says Joseph.
Jim Hare, Distinguished VP Analyst at Gartner, believes Microsoft needs to get three things right for Copilot to scale: clear ROI, seamless integration, and pricing aligned with value. Businesses, he notes, will only adopt it widely if it demonstrably saves time or increases output in measurable ways.
“Copilot must feel like a natural extension of tools people already use (like Office, Windows, and Azure), not an add-on that disrupts workflows,” he adds.
OpenAI Woes
Microsoft was one of OpenAI’s earliest investors, investing over $13 billion in the company over the years, and building a close and exclusive alliance. It holds a 27% stake in the company, valued at $135 billion. In the October-December 2025 quarter alone, Microsoft earned $7.6 billion from the revenue-share agreement with OpenAI, according to TechCrunch.
However, the relationship has been rocky, of late, especially after OpenAI signed several agreements with Amazon, including, and most importantly, with AWS, which will become the exclusive third-party cloud provider for OpenAI Frontier, its new enterprise platform.
This is part of Amazon’s $50 billion investment in OpenAI as part of its latest funding round, including $15 billion upfront and an additional $35 billion tied to milestones.
However, the deal did not sit well with Microsoft, which is reportedly now weighing legal action against OpenAI.
At the centre of the dispute is whether OpenAI can offer Frontier through AWS without breaching contractual terms that require access to OpenAI’s models to be routed through Microsoft’s Azure cloud.
Microsoft believes the arrangement may violate the “spirit”, if not the letter, of its agreement with OpenAI.
“It looks less like a clean breakup and more like things getting a bit messy. OpenAI branching out to AWS plus reports that Microsoft might even sue over the deal for breaking Azure exclusivity, shows the relationship is definitely under strain,” Hare observes.
However, he adds that the companies are still pretty tied together through money, products, and shared tech. “Instead of fully sticking together or splitting apart, it’s probably heading toward this awkward middle ground: competing in some areas (like cloud infrastructure) while still depending on each other in others,” he notes.
In a joint blog post published on February 26, Microsoft and OpenAI clarified that Azure will remain the exclusive cloud provider for stateless OpenAI APIs that deliver access to the company’s models and intellectual property. This means that even if OpenAI partners with AWS or other cloud providers for workload, any stateless API calls to its models must continue to run on Azure infrastructure.
However, it won’t stop the two companies from drifting apart, and Microsoft is already looking elsewhere.
Microsoft has invested up to $5 billion in Anthropic, announced in November last year, as part of a broader strategic push that also includes NVIDIA’s commitment of up to $10 billion. It recently launched Copilot Cowork, a feature in Microsoft 365 Copilot and powered by Claude that allows its AI assistant to execute tasks across workplace applications.
It also backed the Dario Amodei-led company amid Anthropic’s troubles with the Pentagon, which classified it as a “supply-chain risk” over the AI company’s refusal to engage its tech in mass surveillance or developing automated weapons.
In an amicus brief, Microsoft, along with Google, Amazon, and Apple, expressed concerns about the US government’s retaliation. The government’s behaviour could cause “broad negative ramifications for the entire technology sector”, Microsoft warned.
At the same time, OpenAI is no longer solely dependent on Microsoft. It recently raised $110 billion in a funding round led by Amazon, alongside NVIDIA and SoftBank.
While the intimate relationship between the companies has now evolved into an open arrangement. This was the case after October 2025, when the companies restructured their relationship and eliminated Microsoft's right of first refusal to be OpenAI's exclusive compute provider in light of the GPU shortage.
For now, the tensions haven’t boiled over but remain a major sticking point to Microsoft’s AI ambitions.
Your reaction
Discussion
What's your take on this story?
No comments yet — be the first to share your view.