{"slug": "just-wrapped-our-earnings-call-it-was-a-very-strong-close-to-what-was-a-record", "title": "Just wrapped our earnings call. It was a very strong close to what was a record fiscal year for Microsoft. [Read more]", "summary": "Microsoft reported record fiscal year revenue of $331 billion, up 18%, with Microsoft Cloud revenue reaching $214 billion (up 27%) and Azure surpassing $100 billion (up 41%), according to CEO Satya Nadella in a LinkedIn post. Nadella highlighted the company's new model system separating harness, context, memory, and action space from any one model family, and announced a Copilot 'super app' coming this quarter. Critics noted Microsoft's layoffs of roughly 14% of its 2023 workforce amid record profits, questioning the human cost and the loss of institutional knowledge.", "body_md": "Just wrapped our earnings call. It was a very strong close to what was a record fiscal year for Microsoft.\n· Annual revenue: $331B, +18%\n· MS Cloud: $214B, + 27%\n· And Azure: $100B+, +41%\nAnd even bigger opportunity ahead!\nI wanted to share some more perspective on two areas of focus for us:\nFirst, 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.\nAnd it’s not just about cost. It also has the added benefit of business continuity and resilience because every model is substitutable.\nThis is the system we are using in our products, with great results. And we are making it available for our customers via Foundry too.\nYou can read more about this from **Mustafa Suleyman** here: [https://lnkd.in/gQwfAbuc](https://www.linkedin.com/redir/redirect?url=https%3A%2F%2Flnkd%2Ein%2FgQwfAbuc&urlhash=fCIe&trk=public_post-text)\nSecond, when it comes to Copilot, we are innovating rapidly, from chat to Cowork to Autopilots.\nWe have also been super focused on improving the quality and performance of Copilot. Over the last three quarters, user satisfaction scores have doubled. And this quarter alone, we cut latency by 25 percent.\nCopilot usage intensity speaks for itself:\n· Number of conversations per user nearly doubled year over year\n· Average weekly engagement on par with Outlook and Teams\n· Number of customers with over 50K seats up 7X Y/Y\n· Customers deploying Copilot to majority of information workers, up nearly 75% QoQ\nAnd this quarter we will bring all our Copilot experiences together in one “super app” spanning both our consumer and commercial experiences. I’m greatly looking forward to this!\nA big Thank you to our employees and to our customers for continuing to put their trust in us. Read more here: [https://lnkd.in/gR3xdrUb](https://www.linkedin.com/redir/redirect?url=https%3A%2F%2Flnkd%2Ein%2FgR3xdrUb&urlhash=hWVO&trk=public_post-text)\n\nNot just going to be one of the AI hype minions in these comments. Perhaps the missing metric is the human cost of all that efficiency. Microsoft has announced layoffs equivalent to roughly 14% of its 2023 workforce over the past few years, even as it reports record revenue and AI growth. That is not the same as a 14% net reduction in headcount—but it is still a significant amount of disruption. The post talks about the importance of context and memory in AI systems. Organizations need context and memory too, and much of it lives in experienced employees and subject-matter experts. Are you measuring the expertise lost, the work that must be relearned, the trust damaged and the downstream quality effects of repeatedly removing people while celebrating record results? Models may be substitutable. Institutional knowledge often isn’t.\n\nOf those revenue figures, how much was from the salaries of all the people you laid off and their benefits vs how much was actual innovation or strategic (laying people off doesn't count a strategy in this question)?\n\nThe point about separating the harness, context, memory, and action space from any one model family really stands out. As models become increasingly substitutable, the enduring advantage moves to the engineering platform that provides trusted context, governance, evaluation, and observability for AI agents to operate reliably at scale.\n\n[Frank Howard](https://www.linkedin.com/in/frankdhoward?trk=public_post_comment_actor-name)6d\n\nRecord earnings are great, but let’s not ignore the elephant in the room: most innovations in AI are only as good as the data they’re trained on. If that data is flawed, we're just polishing a turd.\n\nRecord revenue again this quarter, which is great, but it comes right after another 4,800 layoffs in July on top of 15,000+ cuts in 2025. The people who actually built this growth aren’t the ones benefiting from it. Also kind of wild how many of the comments here are written by bots or AI.\n\n[Avi Sharma](https://uk.linkedin.com/in/avi-sharma?trk=public_post_comment_actor-name)6d\n\nI keep reading \"every model is substitutable\" in these comments, and it's true right up until you remember customers don't buy models, they buy trust. Swap the engine all you like, the relationship underneath is the one thing you can't reprovision when it breaks.\n\nThe strategic shift is clear: durable advantage will come from owning the orchestration layer, not depending on any single model.\n\nAn important architectural shift here is that the model is no longer becoming the center of the AI system. The intelligence increasingly resides in the enterprise cognitive loop—Sense → Understand → Decide → Action → Learn. Microsoft Graph and connectors continuously sense enterprise signals. Copilot and semantic context ground understanding. Agent Framework and Foundry orchestrate decisions across multiple models. Power Platform and enterprise APIs execute actions. Memory and telemetry close the learning loop. This is what makes true model substitutability possible. The foundation model becomes an interchangeable component, while the enterprise operating system remains persistent, continuously learning and adapting. I believe this is the direction AI-native enterprises will ultimately evolve toward—not AI built around a model, but AI built around a continuously adaptive enterprise intelligence system.\n\n\"Every model is substitutable\" is the line enterprise architects will quote for the next two years. The objection I hear most in EMEA boardrooms isn't \"which model is best.\" It's \"what happens to my system when the model changes underneath it.\" Separating harness, context, memory and action space from the model family turns that from an existential risk into a configuration decision. Portability is the feature. Cost is the consequence.\n\n[See more comments](https://www.linkedin.com/signup/cold-join?session_redirect=https%3A%2F%2Fwww%2Elinkedin%2Ecom%2Fposts%2Fsatyanadella_just-wrapped-our-earnings-call-it-was-a-activity-7488369807969370112-3kA1&trk=public_post_see-more-comments)", "url": "https://wpnews.pro/news/just-wrapped-our-earnings-call-it-was-a-very-strong-close-to-what-was-a-record", "canonical_source": "https://www.linkedin.com/feed/update/urn:li:activity:7488369807969370112/", "published_at": "2026-07-30 18:35:00+00:00", "updated_at": "2026-08-05 14:51:50.184677+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-products", "ai-infrastructure", "ai-agents", "ai-ethics"], "entities": ["Microsoft", "Satya Nadella", "Mustafa Suleyman", "Azure", "Microsoft Cloud", "Copilot", "Foundry"], "alternates": {"html": "https://wpnews.pro/news/just-wrapped-our-earnings-call-it-was-a-very-strong-close-to-what-was-a-record", "markdown": "https://wpnews.pro/news/just-wrapped-our-earnings-call-it-was-a-very-strong-close-to-what-was-a-record.md", "text": "https://wpnews.pro/news/just-wrapped-our-earnings-call-it-was-a-very-strong-close-to-what-was-a-record.txt", "jsonld": "https://wpnews.pro/news/just-wrapped-our-earnings-call-it-was-a-very-strong-close-to-what-was-a-record.jsonld"}}