Microsoft has lowered sales growth targets for its AI agent products after many salespeople missed their quotas in the fiscal year ending in June, according to a report Wednesday from The Information. The adjustment is reportedly unusual for Microsoft, and it comes after the company missed a number of ambitious sales goals for its AI offerings.
AI agents are specialized implementations of AI language models designed to perform multistep tasks autonomously rather than simply responding to single prompts. So-called “agentic” features have been central to Microsoft’s 2025 sales pitch: At its Build conference in May, the company declared that it has entered “the era of AI agents.”
The company has promised customers that agents could automate complex tasks, such as generating dashboards from sales data or writing customer reports. At its Ignite conference in November, Microsoft announced new features like Word, Excel, and PowerPoint agents in Microsoft 365 Copilot, along with tools for building and deploying agents through Azure AI Foundry and Copilot Studio. But as the year draws to a close, that promise has proven harder to deliver than the company expected.
On Tuesday, Microsoft and Nvidia announced plans to invest in Anthropic under a new partnership that includes a $30 billion commitment by the Claude maker to use Microsoft’s cloud services. Nvidia will commit up to $10 billion to Anthropic and Microsoft up to $5 billion, with both companies investing in Anthropic’s next funding round.
The deal brings together two companies that have backed OpenAI and connects them more closely to one of the ChatGPT maker’s main competitors. Microsoft CEO Satya Nadella said in a video that OpenAI “remains a critical partner,” while adding that the companies will increasingly be customers of each other.
“We will use Anthropic models, they will use our infrastructure, and we’ll go to market together,” Nadella said.
Anthropic CEO Dario Amodei, Microsoft CEO Satya Nadella, and Nvidia CEO Jensen Huang discuss the new partnership.
The frenzy of AI deals and cloud partnerships reached another zenith Tuesday morning as Microsoft, Nvidia, and Anthropic announced a surprise alliance that includes a $5 billion investment by Microsoft in Anthropic — which, in turn, committed to spend at least $30 billion on Microsoft’s Azure cloud platform.
Nvidia, meanwhile, committed to invest up to $10 billion in Anthropic to ensure the Claude maker’s frontier models are optimized for its next-generation Grace Blackwell and Vera Rubin chips.
The deal reflects growing moves by major AI players to collaborate across the industry in an effort to build and expand capacity and access to next-generation AI models. Microsoft recently renegotiated its partnership with OpenAI and has been increasingly partnering with others in the industry.
Anthropic has been closely tied to Amazon, which has committed to invest a total of $8 billion in the startup. Anthropic says in a post that Amazon remains its “primary cloud provider and training partner” for AI models. We’ve contacted Amazon for comment on the news.
OpenAI, for its part, recently announced a seven-year, $38 billion agreement with Amazon to expand its AI footprint to the Seattle tech giant’s cloud infrastructure.
Beyond the massive capital flows, the Microsoft-Nvidia-Anthropic partnership expands where enterprise customers can access Anthropic’s technology. According to the announcement, Microsoft customers will be able to use its Foundry platform to access Anthropic’s next-generation frontier models, identified as Claude Sonnet 4.5, Claude Opus 4.1, and Claude Haiku 4.5.
Microsoft also committed to continuing access for Claude across its Copilot family, ensuring the models remain available within GitHub Copilot, Microsoft 365 Copilot, and Copilot Studio.
The news comes as Microsoft holds its big Ignite conference in San Francisco.
Microsoft Azure halted a record 15.72 Tbps DDoS attack from the Aisuru botnet exposing risks created by exposed home devices exploited in large-scale cyber attacks.
An interactive portal at Microsoft’s new Experience Center One grounds visitors in scenes of nature. (GeekWire Photo / Todd Bishop)
[Editor’s Note: Agents of Transformation is an independent GeekWire series and 2026 event, underwritten by Accenture, exploring the people, companies, and ideas behind the rise of AI agents.]
REDMOND, Wash. — If AI were a religion, this would probably qualify as a cathedral.
On the edge of Microsoft’s headquarters, overlooking Lake Bill amid a stand of evergreens, a new four-story building has emerged as a destination for business and tech decision-makers.
Equal parts briefing center, conference hall, and technology showroom, Microsoft’s “Experience Center One” offers a curated glimpse of the future — guided tours through glowing demo rooms where AI manages factory lines, models financial markets, and helps design new drugs.
It’s part of a larger scene playing out across tech. As Microsoft, Google, Amazon and others pour billions into data centers, GPUs, and frontier models, they’re making the case that AI represents not a bubble but a business transformation that’s here to stay.
Microsoft’s new Experience Center One on the company’s Redmond campus was designed by WRNS Studio. (Photo by Jason O’Rear)
As the new center shows, Microsoft’s pitch isn’t just about off-the-shelf AI models or run-of-the-mill chatbots — it’s about custom agentic systems that act on behalf of workers to complete tasks across a variety of tools and data sources.
That idea runs through nearly everything inside the facility, a glass-encased building featuring an elevated garden in a soaring open-air atrium, just across from Microsoft’s new executive offices on its revamped East Campus.
Experience Center One highlights what Microsoft calls “frontier firms” — ambitious companies using AI to push their operations to the edge of what’s possible in their industries.
Agentic AI is “fast becoming the next defining chapter of a frontier organization,” said Alysa Taylor, Microsoft chief marketing officer for Commercial Cloud and AI, in an interview.
The underlying message is clear: get on board or risk falling behind, both competitively and financially. A new IDC study, commissioned by Microsoft, finds both opportunity in spending big and risk in not being bold enough. Companies integrating AI across an average of seven business functions are realizing a return on investment of 2.84 times, it says. In contrast, “laggards” are seeing returns of 0.84 times — basically losing money on their initial spend.
The divide extends to revenue, too: 88% of frontier firms report top-line growth from their AI initiatives, compared to just 23% of laggards, according to the IDC study.
And hey, somebody has to foot the bill for those multi-billion-dollar AI superfactories.
For this second installment in ourAgents of Transformation series, GeekWire visited the new Microsoft facility to see first-hand how the company is presenting its vision of the future. Here are some of the takeaways from the sampling of demos we saw.
These are not off-the-shelf solutions. Each demo reflects a custom deployment built with a major customer, showing how AI tools can be tailored to specific business problems.
Collin Vandament of Microsoft demonstrates a BlackRock investment-analysis scenario inside Experience Center One, showing how a custom AI copilot can translate natural-language questions into the firm’s proprietary BQL code during a tour of the new facility in Redmond. (GeekWire Photo / Todd Bishop)
For example, one shows how Microsoft has worked with BlackRock to integrate a custom AI copilot inside the investment firm’s Aladdin platform to help analysts process large volumes of client and market data more efficiently. It helps reduce the manual work of gathering data and points analysts to potential risks sooner than they might have spotted it on their own.
As another example of the customization, the system is trained to translate natural language requests into “BQL,” BlackRock’s proprietary programming language.
This deep level of integration tracks with the findings in the IDC report. It found that 58% of “frontier firms” are already relying on custom-built or fine-tuned solutions rather than generic models. This is expected to accelerate, with 70% planning to move toward customized tools in the next two years to better handle their proprietary data and compliance needs.
“That’s a trend that we’ve seen even in the low-code movement — taking an out-of-the-box solution, extending it, and customizing it,” said Taylor, the Microsoft commercial CMO.
OpenAI integration remains critical for many Microsoft customers. Another demo focused on Microsoft’s work with Ralph Lauren, showing how the “Ask Ralph” assistant interprets a shopper’s intent and recommends full outfits from available inventory.
Like many of the scenarios inside Experience Center One, this experience runs on Microsoft’s Azure OpenAI Service. It’s a reminder that Microsoft’s partnership with OpenAI — renewed and expanded in recent months — is still a key driver of commercial demand for the tech giant, even as both companies increasingly work with other industry partners.
Teams of agents are starting to redefine industrial work. The clearest example of this was a digital twin simulation from Mercedes-Benz — essentially a virtual version of a factory that lets engineers anticipate and diagnose issues without stopping real production.
The demo begins with a production alert triggered by a drop in efficiency. In a real plant, tracking down the cause (something as small as a slight angle change in a screw) might take a team of specialists days of sorting through machine logs and sensor data.
In Microsoft’s version, a human manager simply asks the system to diagnose what’s causing the problem, through a natural language interface. That question triggers a set of AI agents, each with a specific role: one pulls the right data, another retrieves machine logs, and a third interprets what it all means in plain language.
Within about 15 minutes, the system produces a clear explanation of the likely cause and possible fixes, shortcutting a task that could otherwise stretch across most of a week.
AI is compressing weeks or months of scientific research into days or hours. A demo focusing on Insilico Medicine’s work with Microsoft showed how AI is starting to significantly collapse the timeline for drug discovery.
The process begins with a “digital researcher” that scans huge amounts of public biomedical data to surface promising disease targets. It’s the kind of work that would otherwise take teams of scientists months of reading and analysis.
Collin Vandament demonstrates an Insilico Medicine drug-discovery scenario inside Experience Center One, where interactive displays visualize how AI models surface potential disease targets and rank candidate molecules. (GeekWire Photo / Todd Bishop)
A second system runs simulated chemistry experiments in the cloud, generating and ranking potential molecules that might bind to those targets. These simulations can be completed in a matter of days, or less, replacing weeks or even months of traditional laboratory work.
The demo follows a real example: Insilico used this workflow to identify a potential target for a lung disease and design molecules that could affect it. The company then synthesized dozens of these AI-generated compounds in the lab. One of them is now in Phase 2a human trials.
That’s a small sampling of the demos inside Microsoft’s Experience Center One.During our tour, we walked past displays for other major brands, including more iconic U.S. companies, but not everyone was willing to have the media spotlight cast upon their projects. As a condition of access, we agreed to stick to the examples cleared for public release.
Of course, the demos are carefully curated, and it remains to be seen how broadly companies will deploy these kinds of systems in their real-world operations.
The open-air atrium at Microsoft’s Experience Center One brings natural light into the building. (GeekWire Photo / Todd Bishop)
In many ways, the facility is a successor to Microsoft’s longtime Executive Briefing Center and conference facility, which remains in use a short walk away on the Redmond campus.
Trevor Noah with Hadi Partovi of Code.org during an October event at Experience Center One. (GeekWire Photo / Taylor Soper)
It’s closed to the public, invite-only. Employees can request access to visit.
Visitors arrive via a circular drive, with a plaque at the entrance dedicating the building to John Thompson, the former Microsoft board chair who led the search process that resulted in Satya Nadella’s appointment as CEO. There are private briefing suites on the upper floors, and a full cafe on the second. The building also includes a conference center with three auditoriums.
But perhaps the most distinct feature is the interactive portal. As they leave the demos, visitors walk through an immersive digital corridor with scenes of nature on the virtual walls.
Walking through the tunnel, motion sensors track their movement, causing digital leaves and particles on the wall-sized screens to swirl and flow in their wake.
The audio consists of nature sounds (birds, wind, and rustling trees) that were recorded locally in the Redmond and Sammamish area. And in a fittingly Pacific Northwest touch, the visual display is connected to a weather API. If it has been raining outside (as it often has been recently) the digital environment inside the tunnel turns rainy, too.
It’s meant to be a final moment of grounding — a programmed moment of Zen to help executives decompress and center themselves as they contemplate the frontier ahead.
A major Microsoft outage has disrupted Azure, Microsoft 365, Xbox, and Minecraft worldwide after a configuration failure, with services now gradually recovering.
Microsoft reported fiscal first-quarter revenue and profits ahead of analysts’ expectations on Wednesday, with Azure revenue growth climbing to 40%.
The earnings report came as the company continued to deal with the lingering effects of a widespread cloud outage that started earlier in the day.
The company’s capital expenditures reached a record $34.9 billion — reflecting its long-term buildout of cloud infrastructure to meet demand for artificial intelligence. That was up from $24.2 billion in Q4. Microsoft had projected capital spending of more than $30 billion for Q1.
Along with that unprecedented buildout, Microsoft sought to address investor concerns about a potential AI bubble, by highlighting its commercial remaining performance obligation (RPO), a measure of future contracted revenue. That backlog grew 51% year-over-year to $392 billion.
The company also disclosed for the first time that this RPO has a weighted average duration of roughly two years, a move intended to show investors that its record capital spending is supported by strong, long-term customer demand.
Revenue was $77.7 billion for the quarter ended Sept. 30, Microsoft’s first quarter of fiscal 2026. That was up 18%, and compared with average analyst expectations of $75.39 billion. The company said the result was driven by strong demand for cloud and AI services.
Profits were $27.7 billion, or $3.72 per share, beating expectations of $3.66 per share.
Earlier Wednesday, an Azure cloud services outage disrupted operations for customers worldwide including Alaska Airlines, Xbox users and Microsoft 365 subscribers. Microsoft reported as of early afternoon that it was rolling back the faulty configuration and that customers should see improvements.
Microsoft stock was down by about 3% in after-hours trading. The company’s market value reached $4 trillion after the announcement of its new OpenAI deal on Tuesday morning.
An outage on Microsoft’s Azure cloud services Wednesday morning disrupted operations for customers worldwide including Alaska Airlines, Xbox users and Microsoft 365 subscribers.
The incident came just ahead of Microsoft’s quarterly earnings call today and follows last week’s outage at Amazon Web Services and a failure of Alaska Airlines’ own data center technology.
The latest outage struck at 9 a.m. PT, according to Microsoft, when the system “began experiencing Azure Front Door (AFD) issues resulting in a loss of availability of some services. We suspect that an inadvertent configuration change as the trigger event for this issue.
“We are taking several concurrent actions: Firstly, where we are blocking all changes to the AFD services, this includes customer configuration changes as well. At the same time, we are rolling back our AFD configuration to our last known good state,” the company stated. “As we rollback we want to ensure that the problematic configuration doesn’t re-initiate upon recovery.”
Alaska Airlines posted on X at 10:33 a.m., explaining that the Azure outage was disrupting systems including their website function. Passengers flying on Alaska and Hawaiian airlines who were unable to check-in online were directed to airline agents to receive their boarding passes.
“We apologize for the inconvenience and appreciate your patience as we navigate this issue,” the post said.
Microsoft did not indicate when the outage would be resolved.
“We do not have an ETA for when the rollback will be completed, but we will update this communication within 30 minutes or when we have an update,” the company posted at 10:51 a.m.
UPDATE: At 12:22 p.m. the company shared an update stating it had deployed the “last known good” configuration of the impacted system and customers should start seeing improvements. “[W]e anticipate full mitigation within the next four hours as we continue to recover nodes …. We will provide another update on our progress within two hours, or sooner if warranted,” Microsoft added.
Days after its outage last week, AWS offered a detailed explanation of the event, which was caused by a cascading failure triggered by a rare software bug in one of the company’s most critical systems. The disruption impacted sites and online services around the world.
Alaska Airlines attributed its recent outage to a failure at its primary data center. The company operates a hybrid infrastructure, blending its own data centers with third-party cloud platforms. The incident disrupted travel for more than 49,000 passengers.
Do you remember the good old days when you could march right up to the server and give it a good whack on the side? You know, the old ones-two fingers-across-the-top. It didn't work a lot of the time, but when it did, it worked like magic.
AzureStrike is a red team toolkit for attacking Azure Active Directory, enabling reconnaissance, credential abuse, and persistence in cloud environments.
Minimizing who can access your data and when is one of the cornerstones of cybersecurity as it helps to decrease the chance of sensitive information falling into the hands of a malicious actor. It also protects data against being accidentally viewed (or even inadvertently leaked!) by an authorized user.
Because privileged user accounts hold higher levels of access than other user accounts, they need to be monitored more closely. PIM is a service in Azure Active Directory that allows you to restrict access in a variety of cool ways, from making it time-bound to implementing just-in-time access.
In her exploration of Privileged Identity Management in Azure Active Directory, Paula covers:
Assigning roles
Adding assignments
Giving global administrative rights to a user
Configuring limited time access that expires after a specified time
How to activate a role and monitor it using Assigned Admins
Holiday time is approaching and we know that everyone loves to receive gifts! Especially at CQURE, the idea of sharing is close to us and we would like to invite you to our Great Racoon Giveaway Contest, where you will get a chance to win $3920-worth voucher for any of CQURE Academy Live Courses!
Please click on the below banner to find out more about the contest:
Identity Protection is a security feature in Azure Active Directory that helps to prevent, detect, and remediate identity risk in an organization. Using multiple detections, it monitors every login for identity compromise, sorting sign-ins into three categories of risk: low, medium, and high.
These risk ratings can be used to create automated user risk policies that balance employee productivity with corporate security. For example, multi-factor authentication can be set as a requirement for a sign-in that is high-risk.
Join Paula as she reviews the different policies in Azure’s Identity Protection (User Risk, Sign-in Risk, and MFA Registration) and explains how to:
Select which users you want to include in the policy
Exclude specific users (such as your ‘break-glass’ account so that you cannot be accidentally logged out of Azure Active Directory)
Specify risk levels as high, medium, or low in the User Risk section
Block access or allow access but require a password change in the Access section
Activate and enforce a policy that you have set up and configured
Paula shows how to monitor your organization for risky users and risky sign-ins in the Report section of Azure’s Identity Protection dashboard and takes you through how to delete the conditional access policies you create.
Discover what happens when a log-in to an organization’s Microsoft Office portal from a Tor browser is flagged as “something strange” by Azure AD’s Identity Protection. You’ll also learn how to mark identity as compromised if, for example, sign-ins have been made in two completely different locations using that identity.
Paula covers identity security from the perspectives of both the administrator and the user, giving a clear view of the steps an employee must take when their account has been identified as risky.
With this identity security lesson under your belt, you’ll be able to intelligently react to potentially dangerous situations. Take a stroll around the CQURE Academy blog now for more Azure Active Directory security tips including ‘8 things to avoid’ in Azure AD.
Holiday time is approaching and we know that everyone loves to receive gifts! Especially at CQURE, the idea of sharing is close to us and we would like to invite you to our Great Racoon Giveaway Contest, where you will get a chance to win $3920-worth voucher for any of CQURE Academy Live Courses!
Please click on the below banner to find out more about the contest:
Regulating access to your company’s files, systems, and applications cuts the risk of your data falling into the hands of hackers, threat actors and thieves.
While standard privilege management stops at ID-based authentication, conditional access in Azure Active Directory gives greater flexibility and control by allowing remote connections only when certain conditions are met.
Using conditional access, an administrator can regulate access by user location, device type, the kind of application or file being used and more. To achieve this, the administrator creates an Azure Active Directory security policy that specifies which condition(s) must be met for access to be allowed.
In this back-to-basics CQURE Hacks episode, Paula J demonstrates how to create secure conditional access policies and monitor access in the Azure Active Directory.
>>> Controlling access by a user’s IP address
o Add the IP range’s location
o Define the range to be assigned to the policy
o Name the policy e.g., ‘Corporate IP range’
o Specify the trusted IP addresses related to the location
>>> Controlling access by the kind of user or group, e.g., corporate only
o Create a new policy
o In conditions, specify login from corporate IP addresses
o Exclude sign-ins from other users and groups
>>> Controlling access by location
o A demonstration using the United States and Poland as examples
>>> Creating emergency access accounts known as “break glass accounts” to prevent yourself being accidentally locked out of your Azure Active Directory
>>> More ways to regulate access
o Blocking access
o Enforcing multifactor authentication
o Session controls
>>> Final steps
o Turning on policies
o Testing polices
o Monitoring user access via the dashboard
After you’ve set up conditional access in Azure Directory, browse our blog to discover more clever ways to secure your data.
Login
