‘If a firm is a learning machine, they need their own learning machine. Models are an input, not some extraction of the knowledge of the enterprise,’ Microsoft CEO Satya Nadella said on the company’s fourth fiscal quarter earnings call.
Microsoft Chairman and CEO Satya Nadella touted his company’s products and services as the best interaction layers for any variety of artificial intelligence models while maintaining control over their data and intellectual property.
Nadella’s goal is enabling customers to “be in control of their own destiny in terms of what I describe as building their human capital and their token capital,” the CEO said on Wednesday’s fourth fiscal quarter earnings call for the Redmond, Wash.-based technology giant.
He predicted that more and more AI users will choose vendors that prove they are not using company knowledge to benefit themselves.
“If a firm is a learning machine, they need their own learning machine,” Nadella said. “The models are an input, not some extraction of the knowledge of the enterprise.”
Microsoft reported results for its fourth fiscal quarter on the call, covering the three months ended June 30.
[RELATED: Microsoft’s Nicole Dezen Unveils New Partner Strategy Built Around AI Transformation
Zac Paulson, vice president of technology at Fargo, N.D.-based Microsoft solution provider ABM Technology Group—a member of CRN’s 2026 MSP 500—told CRN in a recent interview that ABM’s security offerings have seen high demand lately, with customers concerned about the volume of attacks and new attack surfaces in the AI era.
“AI has notched up the threats, so people are more cognizant about what to worry about,” Paulson said.
Paulson said he has also been impressed with Microsoft’s ongoing advancements in its Copilot AI tool and that Microsoft’s communications application Teams has been growing in adoption as a phone system in a non-AI business opportunity and speaking to Microsoft’s value-add as a one-stop productivity vendor.
“It is emotionally taxing to have 27 different places you have to go [to for IT products],” Paulson said. “We can make the Microsoft ecosystem the one place—which is always where Microsoft has shined. Their Office products, everything works.”
Microsoft Pushes Multimodel AI Strategy
One of Microsoft’s major differentiators in the AI race is through separating AI harnesses from the models to differentiate its architecture, Nadella said. The harnesses keep memory and context external and allow users to swap in any model they want at any given time, perhaps a nod toward the commoditizing nature of AI models.
While some AI vendors advocate open-source and small language models as a way to reduce costs, Nadella said enterprises shouldn’t avoid frontier models outright.
“There’s no reason not to [use large models],” he said. “You also can use multiples of them” in addition to low-cost models and client-made models.
“We want to democratize that design pattern so that every enterprise can use it,” the CEO said.
Nadella pointed to the recent incident of OpenAI models autonomously compromising model platform Hugging Face as an example of why users shouldn’t depend on one model.
“You will maybe need multiple models to even remediate some challenges that get caused by one model,” Nadella said. “You can’t be subject to the refusals of one model.”
Microsoft’s AI model catalog now includes more than 11,000 models, including those from OpenAI, Anthropic, Mistral, Elon Musk’s xAI and first-party ones.
The number of customers using models from multiple providers has grown fivefold since the start of the year, Nadella said.
Although Nadella spoke to the importance of leveraging models by multiple makers, he still pointed to efficiencies users can achieve by leveraging Microsoft models with Microsoft products and services.
“We are building a new model system where the harness, context, memory and action space are separate from any one model family, thereby moving the frontier on the cost to outcome curve,” Nadella said on the call. “It’s not just about cost. It also has the added benefit of business continuity and resilience because every model is substitutable.”
Nadella said that millions of developers, for example, have used Microsoft AI’s MAI-Code-1-Flash model on GitHub Copilot to achieve higher acceptance rates and 10 percent lower median token usage while still having access to OpenAI and Anthropic models.
In another example, users of Microsoft’s classic Excel spreadsheet application can use Microsoft’s MAI-Code-1-Flash model for a comparable quality to OpenAI’s GPT-5.6 model for common tasks at lower costs.
In security, the MAI-Cyber-1-Flash model has seen better performance than Anthropic’s Mythos model at half the cost when combined with Microsoft’s multi-agent security harness, the CEO said.
“For whatever reason, if a given model goes away then you can’t be left high and dry—you need to be able to still continue your cyber operations,” Nadella said. “It is cost and resilience [that are] both important criteria. And that’s what we are trying to build in, whether it’s in code, whether it’s in cyber, whether it is in knowledge work.”
Copilot Tops 30 Million Paid Seats
As Nadella traditionally does, he shared a variety of product usage milestones to show Microsoft’s growth across different categories.
In database products, Microsoft saw PostgreSQL revenue grow 55 percent year on year, accelerating for the third consecutive quarter, the CEO said.
The number of PostgreSQL customers also using Microsoft’s Foundry platform for building, deploying and governing generative AI applications and autonomous agents increased 80 percent.
Microsoft now has more than 40,000 paid Foundry customers, up more than 60 percent year over year. Revenue more than doubled year over year, with the number of Foundry customers at a 1 trillion token annualized run rate increasing fourfold year over year, the CEO said.
More than 17,000 customers use Foundry with Microsoft’s Fabric data storage and processing service, up 60 percent year over year. Tens of thousands of customers, including nearly 90 percent of the Fortune 500, are grounding agents in enterprise context with Microsoft Foundry, Fabric and the Work IQ organizational intelligence and context layer behind Microsoft 365 Copilot.
Other product and service milestones include:
- The vendor’s Rayfin backend-as-a-service (BaaS) platform and agent-first software development kit (SDK) has more than 2,500 customers.
- Only two months after the launch of Agent 365, the agent control plane has nearly 40 million agents registered across tens of thousands of companies.
- Microsoft now has more than 30 million paid Microsoft 365 Copilot seats, with net seat adds more than doubling quarter over quarter.
- Over the last three quarters, user satisfaction scores for Copilot have doubled and are at an all-time high.
- During the quarter, Microsoft cut Copilot latency by 25 percent.
- The number of conversations per user nearly doubled year over year.
- Average weekly engagement is on par with Outlook and Teams.
- The time it takes customers to reach what Microsoft considers “high usage”—about 80 percent monthly active use across a user base—has fallen from months to days over the past year.
- The number of customers with more than 50,000 Copilot seats increased more than sevenfold year over year.
- The number of enterprise customers deploying Copilot to the majority of their information workers grew nearly 75 percent quarter over quarter.
- Two months after the launch of the E7 license, hundreds of enterprise customers have purchased millions of seats.
- GitHub Copilot now has 50 million users.
- Copilot revenue accelerated more than 60 percent quarter over quarter.
- GitHub now has 225 million users, including more than 90 percent of the Fortune 500.
- Purview has audited more than 15 billion Copilot interactions to meet compliance obligations, more than a quadruple percent year over year.
- Microsoft is on pace to automate more than 100 million health-care patient encounters this calendar year.
CRN 2026 Solution Provider 500 members Infosys, Tata Consultancy Services (TCS) and Wipro received acknowledgement from Nadella on the call for each purchasing 60,000 or more Copilot seats.
The new Microsoft Frontier Co. forward-deployed engineering (FDE) organization of 6,000 embedded industry and engineering experts so far has completed 330 projects across 164 customers, Nadella said.
The CEO also held up Microsoft’s emerging licenses plus consumption pricing model as a powerful total addressable market amplifier, with Copilot leveraging both pricing models.
The vendor has been moving more products and services to a consumption or mixed pricing model. Thousands of customers are paying for and actively using the usage-based billing option in the Copilot Cowork agentic AI system for multistep projects and workflows, as one example, Nadella said on the call.
Dynamics 365’s usage-based credit consumption in the customer service category is up fourfold quarter over quarter. GitHub Copilot’s usage-based billing introduced in the quarter did not stop business and enterprise seat growth and produced consumption revenue, Nadella said.
Microsoft’s newly unveiled Perception agentic security system will also be charged by consumption, Nadella said.
Azure Revenue Surpasses $100 Billion
Microsoft brought in $90 billion for the quarter, up 17 percent year on year ignoring foreign exchange.
Azure revenue surpassed $100 billion for the first time during the quarter. Microsoft 365 Copilot surpassed 30 million paid seats.
Microsoft Cloud brought in revenue of $59.3 billion, up 27 percent year on year. Microsoft still has a backlog, or commercial remaining performance obligation (cRPO) of $678 billion, up 84 percent.
In a notable nod to the diversity of the client mix in that backlog, Microsoft CFO Amy Hood said on the call that all sequential cRPO growth was driven by commitments from customers outside frontier model companies like OpenAI and Anthropic. RPO increased 25 percent when excluding OpenAI.
The vendor’s operating income was $40.6 billion, up 18 percent. Net income came in at $35.8 billion, up 31 percent year on year, using GAAP.
Without using GAAP, net income came in at $35.3 billion and increased 22 percent, according to the vendor. Microsoft’s GAAP results include effects from its OpenAI investment while non-GAAP excludes those effects.
Microsoft acknowledged a $3.2 billion discrete gain from its investment in OpenAI competitor Anthropic. The vendor also paid less than expected for its voluntary retirement program for employees, but Hood acknowledged that Microsoft’s head count fell 2 percent year on year.
Investing in AI infrastructure was a factor in Microsoft’s company gross margin percentage coming down year on year to 67 percent. A greater sales shift to Azure was another factor.
Microsoft spent $41 billion in capital expenditure during the quarter, with the growing cost of data center components—which has proven disruptive to the channel overall—a factor.
Hood pointed to Microsoft’s supply chain prowess as a hyperscaler as helpful in the high-priced component environment and indicated that the high prices have helped with conversations around migrating clients to the cloud to avoid buying servers on-premises.
“The cloud still provides a great ROI in those types of situations,” Hood said. “For the long term, you want to have pricing work for customers and for you, and so we’re trying to stay focused on that as well.”
Microsoft Segments In Detail
In Microsoft’s productivity and business processes (PBP) segment, the vendor saw $37.8 billion in revenue, up 14 percent year on year.
In that segment, Microsoft 365 Commercial cloud revenue grew about 14 percent year on year. The consumer cloud grew 22 percent ignoring foreign exchange. Dynamics 365 revenue grew 12 percent. Paid M365 commercial seats grew 6 percent year over year. M365 commercial products revenue increased 19 percent, ahead of expectations, Hood said. She credited the growth to large, long-duration M365 contracts that resulted in higher in-period revenue recognition from the Windows commercial on-premises business.
This segment also brings in revenue from Microsoft social media network LinkedIn.
Microsoft’s intelligent cloud (IC) segment brought in $39.3 billion for the quarter, up 31 percent year on year ignoring foreign exchange. This segment includes Azure and other cloud services revenue, which grew 43 percent.
Microsoft’s more personal computing (MPC) segment brought in $12.9 billion, down 5 percent year on year ignoring foreign exchange.
In that segment, Windows OEM and devices revenue fell 7 percent year on year. Hood said that the fall came in part from lower PC market demand and a high prior-year comparable from Windows 10 end of support at the time.
Those results still beat her expectations as OEM and channel partners built inventory to get ahead of increasing component prices, she said.
This segment also includes revenue from Xbox and search advertising.
Full-Year Results, Microsoft’s Fiscal 2027 Outlook
For the entire fiscal year, Microsoft reported $331.8 billion in revenue, up 16 percent ignoring foreign exchange.
Microsoft reported $155.2 billion in operating income, up 19 percent year on year ignoring foreign exchange. Net income using GAAP grew 31 percent to $133.7 billion. Without GAAP—and without OpenAI impacts—it grew 20 percent ignoring foreign exchange.
For the full year, Microsoft’s cloud revenue surpassed $214 billion, with nearly 90 percent of that from customers outside frontier model companies like OpenAI and Anthropic. Cloud’s gross margin percentage was better than expected at 65 percent but down year over year due in part to AI infrastructure investment, Hood said.
Looking ahead, Microsoft should grow revenue at least $89.85 billion, up at least 16 percent year on year.
Hood still expects M365 commercial product and server product revenue to decline in the mid-single digits for the full fiscal year due in part to lapping higher transactional purchasing from the timing of product launches.
She expects lower PC market demand and higher device pricing from component costs to lead to revenue decline in the high teens for the fiscal year for that business.
Microsoft expects the productivity and business processes segment to bring in revenue of $36.7 billion to $37 billion, or growth of 11 percent to 12 percent.
Hood said to expect M365 commercial cloud growth of about 16 percent ignoring foreign exchange. Sequential growth from Microsoft Copilot, E5 and E7 faces some mitigation from lower average revenue per user (ARPU) new seat adds in frontline worker and small- and medium-business products—which are sold by Microsoft solution providers.
The vendor does expect to see acceleration in M365 commercial cloud revenue growth through this fiscal year with premium license momentum and increased monetization from adding usage-based billing products alongside per-seat licensing, Hood said.
M365 commercial product revenue should grow in the mid-single digits. M365 consumer cloud revenue should grow in the mid-teens due in part to a price increase last year. Dynamics 365 revenue should grow in the low teens, relatively stable quarter over quarter, the CFO said.
The intelligent cloud segment should see revenue of $40.95 billion to $41.25 billion. Azure should grow 45 percent ignoring foreign exchange.
The on-premises server business should see revenue decline in the low- to mid-single digits.
In the more personal computing segment, Microsoft said to expect revenue of $12.2 billion to $12.7 billion. Windows OEM and devices revenue should decline in the low 20s.
“As in prior quarters, the range of potential outcomes remains wider than normal,” Hood said in reference to the impacts of higher component prices.
Microsoft expects to spend more than $50 billion in CapEx as it works to meet AI demand, the CFO said.
Microsoft’s stock rose about 9 percent in after-hours trading Wednesday, trading at about $425 a share.
