ClaimsPeople

Satya Nadella

Satya Nadella is the Chairman and Chief Executive Officer of Microsoft, which he has led as CEO since February 2014.

969 claims collected, 174 researched, 98 resolved. Report of 5 November 2026, method 0.1.0+28da144b.

Scores by kind of claim

Each kind of claim is scored on its own, out of 100: the share that held up, with the range it likely falls in. Fewer resolved claims give a wider range.

  • Factual claims

    Statements about the present or the past, checked against the record: did they hold up?

    83 77 88

    High confidence, 77 factual claims resolved. From 16 documents, 2014 to 2026; 19 documents sampled from 35 candidates found.

    Not scored: 23 unresolvable, 17 checkable only against their own records.

  • Predictions

    Statements about events outside the person's control, checked once they fell due: did they come true?

    Insufficient record: 4 predictions resolved, 5 needed for a score.

    Not scored: 19 unresolvable. 1 more fall due later.

    Does not yet allow for how likely each prediction was when made.

  • Promises

    Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?

    77 65 88

    Medium confidence, 17 promises resolved. From 9 documents, 2014 to 2026; 19 documents sampled from 35 candidates found.

    Not scored: 9 unresolvable, 7 not yet due.

All types combined: 82 (likely 77 to 86, high confidence), from 94 claims: 77 factual claims, 17 promises. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.

Satya Nadella's claims

338 claims, sorted by newest said. Choose a claim to read it in full.
VerdictKindSaidClaimTopic
Mixed Factual 4 Jun 2026 In a Land O'Lakes agent example shown at Build, a 5B-parameter model delivered the same outcome as a 500B-parameter model. Factual · 4 Jun 2026 AI capabilities
Mostly true Factual 4 Jun 2026 Microsoft built SQL Server in partnership with SAP. Factual · 4 Jun 2026 Technology
Not researched Factual 4 Jun 2026 Nadella said they are using the same multi-model harness that they use in GitHub and in security. Factual · 4 Jun 2026 AI capabilities
Unresolvable Factual 4 Jun 2026 Microsoft's free cash flow is currently being allocated to capital return in a sensible way, per Nadella. Factual · 4 Jun 2026 Their own work or organisation
Own records only Factual 4 Jun 2026 Microsoft allocates compute across three buckets: its hyperscale (Azure) business, its own application business, and its own research compute. Factual · 4 Jun 2026 AI compute, hardware and data
Unresolvable Factual 4 Jun 2026 Microsoft gave at most a little more compute to Azure last quarter, and it was supply-constrained. Factual · 4 Jun 2026 AI compute, hardware and data
Mostly false Factual 4 Jun 2026 After moving to the cloud, Microsoft sold many more subscriptions because customers who never bought its servers bought subscriptions. Factual · 4 Jun 2026 Economics and markets
Unresolvable Factual 4 Jun 2026 Microsoft's MAI models are already fully competitive on the frontier with the other general models. Factual · 4 Jun 2026 AI capabilities
Own records only Factual 4 Jun 2026 Microsoft's earlier fine-tuning effort did not work well because it lacked the tools and data collection regime. Factual · 4 Jun 2026 AI capabilities
Own records only Factual 4 Jun 2026 Microsoft's infrastructure business margin dollars are close to exceeding the total margin dollars from its high-margin businesses. Factual · 4 Jun 2026 Economics and markets
Not researched Factual 4 Jun 2026 Microsoft is not selling raw GPU capacity to new AI labs (Neolabs) on Azure. Factual · 4 Jun 2026 AI compute, hardware and data
Own records only Factual 4 Jun 2026 Some of Microsoft's model performance gains come from reverse knowledge distillation (RKLD) using OpenAI IP, plus reinforcement learning on top. Factual · 4 Jun 2026 AI capabilities
Unresolvable Factual 4 Jun 2026 Overall margin dollars from Microsoft's infrastructure business may be higher, even though it is not a higher-margin business; he added that these margin dollars are getting close to being higher than the total margin dollars from Microsoft's high-margin businesses. Factual · 4 Jun 2026 Economics and markets
Unresolvable Factual 4 Jun 2026 Microsoft showed at Build that an efficiently trained reasoning or coding model can hill-climb using a customer's traces and be more token-efficient. Factual · 4 Jun 2026 AI capabilities
Own records only Factual 4 Jun 2026 Microsoft cannot currently add more new AI labs (Neolabs) as Azure customers, does not sell raw GPUs to them, and turns some business away. Factual · 4 Jun 2026 AI compute, hardware and data
Own records only Factual 4 Jun 2026 Microsoft aims not to use distillation during its own hill-climbing of its models, except at the very end. Factual · 4 Jun 2026 AI capabilities
Not researched Factual 4 Jun 2026 Nadella said he had just seen the news and thought Google had just issued equity to fund its build-out. Factual · 4 Jun 2026 Economics and markets
Not researched Factual 4 Jun 2026 Nadella said he thinks an effect similar to the move to the cloud is already happening with agents, and that he sees it on GitHub, M365 and security. Factual · 4 Jun 2026 AI capabilities
Not researched Factual 4 Jun 2026 Microsoft will have a harness independent of the underlying models, and it has a GitHub Copilot harness that is used across Microsoft. Factual · 4 Jun 2026 AI capabilities
Not researched Factual 4 Jun 2026 Until 2032, Microsoft and OpenAI remain mutual customers and Microsoft remains an IP partner of OpenAI. Factual · 4 Jun 2026 Their own work or organisation
Not researched Factual 4 Jun 2026 Inference compute demand has exploded in GitHub and Microsoft 365, requiring Microsoft to fund its own applications. Factual · 4 Jun 2026 AI compute, hardware and data
Mostly true Factual 4 Jun 2026 Microsoft did not raise M365 prices for more than a decade, thanks to software efficiencies on top of hardware. Factual · 4 Jun 2026 Economics and markets
Own records only Factual 2 Jun 2026 Microsoft used the Discovery agentic loop to accelerate its quantum work, compressing years of research into the last year. Factual · 2 Jun 2026 Science and research methods
Not researched Factual 2 Jun 2026 Operations in Majorana 2 take one microsecond, with digital control in the same qubit form factor as Majorana 1. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 PowerPoint Alt Text and Teams Super Resolution use on-device AI running locally. Factual · 2 Jun 2026 Their own work or organisation
Not researched Factual 2 Jun 2026 Microsoft's Autopilots are autonomous, long-running agents with full enterprise compliance that run in the customer's tenant. Factual · 2 Jun 2026 AI capabilities
Not researched Factual 2 Jun 2026 Every Fairwater data center is connected through Microsoft's continent-spanning AI WAN. Factual · 2 Jun 2026 AI compute, hardware and data
Not researched Factual 2 Jun 2026 The second new Windows in-box model is Aion Plan, a planning model providing a local agentic loop. Factual · 2 Jun 2026 AI capabilities
Mostly true Factual 2 Jun 2026 Cowork came as a new way of working that generates artifacts and solves multistep problems, and users assign multistep tasks to it. Factual · 2 Jun 2026 AI capabilities
Not researched Factual 2 Jun 2026 Microsoft announced Majorana 2 on 2 June 2026. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 In the week before June 2, 2026, Microsoft added OpenAI real-time voice models and Claude Opus 4.8 to Foundry. Factual · 2 Jun 2026 Their own work or organisation
Not researched Factual 2 Jun 2026 Windows 365 now has a developer distribution optimized for cloud developer productivity. Factual · 2 Jun 2026 Their own work or organisation
Not researched Factual 2 Jun 2026 Microsoft announced a partnership with Fireworks AI bringing all of its open weight models to Foundry. Factual · 2 Jun 2026 Their own work or organisation
Not researched Factual 2 Jun 2026 Web IQ is model agnostic and MCP native, and plugs into any agent runtime. Factual · 2 Jun 2026 AI capabilities
Not researched Factual 2 Jun 2026 Fairwater is Microsoft's first AI super factory, spanning Georgia and Wisconsin. Factual · 2 Jun 2026 Their own work or organisation
Not researched Factual 2 Jun 2026 Majorana 1 proved out the foundational physics of Microsoft's topological qubit approach, and Majorana 2 begins the engineering-scale phase. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 Agents are using services such as Cosmos DB for their memory, and ChatGPT does so. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 NVIDIA's RTX Spark is a PC system-on-chip combining CPU, GPU and AI capabilities, with unified memory architecture and integrated DRTM. Factual · 2 Jun 2026 AI compute, hardware and data
Not researched Factual 2 Jun 2026 Majorana 2 has a qubit mean lifetime of 20 seconds up to a minute, about 1,000 times higher than Majorana 1. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 OpenClaw runs on Windows leveraging MXC, announced 2 June 2026. Factual · 2 Jun 2026 AI safety and alignment
Not researched Factual 2 Jun 2026 Rayfin is an agent-first SDK that connects agents to a backend as a service, and Microsoft is bringing it to everywhere developers build. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 The machine described, called "a dream machine" and said to be coming in the fall, has 1 petaflop of AI compute, 20 CPU cores and 128 GB of unified memory. Factual · 2 Jun 2026 AI compute, hardware and data
Not researched Factual 2 Jun 2026 On 2 June 2026 Microsoft introduced Autopilots, described as enterprise-grade claws, in Copilot. Factual · 2 Jun 2026 AI capabilities
Not researched Factual 2 Jun 2026 Fairwater's cooling loop is filled once and the data center can operate with effectively zero water consumption. Factual · 2 Jun 2026 AI compute, hardware and data
Not researched Factual 2 Jun 2026 Microsoft is introducing an intelligent Windows terminal with built-in GitHub Copilot. Factual · 2 Jun 2026 Their own work or organisation
Not researched Factual 2 Jun 2026 Nadella said that an agent harness for security that "we built" brings together a hundred agents across frontier and custom models, with the aim of finding exploitable bugs better than any single model does. Factual · 2 Jun 2026 AI safety and alignment
Not researched Factual 2 Jun 2026 Microsoft extended Purview to provide always-on data protection and compliance for AI agents. Factual · 2 Jun 2026 AI safety and alignment
Not researched Factual 2 Jun 2026 Microsoft announced on 2 June 2026 an expansion of the scope of Windows ML and Windows AI to the full install base of GPUs. Factual · 2 Jun 2026 Their own work or organisation
Not researched Factual 2 Jun 2026 Nearly all GitHub metrics, including repo creation, PR activity, API usage and actions, are growing faster because of agentic workflows. Factual · 2 Jun 2026 AI, the economy and work
Not researched Factual 2 Jun 2026 Microsoft introduced Project Solara, a new platform for the agent era, on June 2, 2026. Factual · 2 Jun 2026 Their own work or organisation
Not researched Factual 2 Jun 2026 Agents managed with Microsoft's Agent 365 can be hosted on AWS, GCP or Azure and built with any framework. Factual · 2 Jun 2026 AI safety and alignment
Not researched Factual 2 Jun 2026 Microsoft's Cobalt CPU delivers over 50% better cloud-native performance than Cobalt 100. Factual · 2 Jun 2026 AI compute, hardware and data
Not researched Factual 2 Jun 2026 The DGX Station can run a 1 trillion parameter model locally. Factual · 2 Jun 2026 AI compute, hardware and data
Not researched Factual 2 Jun 2026 For its first 15 or so years Microsoft built commercial cloud infrastructure for heterogeneous enterprise workloads. Factual · 2 Jun 2026 Their own work or organisation
Not researched Factual 2 Jun 2026 As of June 2026, Microsoft Foundry offers over 11,000 models. Factual · 2 Jun 2026 Their own work or organisation
Not researched Factual 2 Jun 2026 From 2 June 2026, Copilot Frontier users can try Scout. Factual · 2 Jun 2026 Their own work or organisation
Not researched Factual 2 Jun 2026 Microsoft announced general availability of Microsoft Discovery on 2 June 2026. Factual · 2 Jun 2026 Science and research methods
Not researched Factual 2 Jun 2026 Foundry provides safety features and guardrails around agentic systems. Factual · 2 Jun 2026 AI safety and alignment
Not researched Factual 2 Jun 2026 MDASH is an agent harness for security that Microsoft built. Factual · 2 Jun 2026 AI safety and alignment
Not researched Factual 2 Jun 2026 Web IQ is built on Microsoft's global infrastructure, which already serves over a billion users, re-architected for LLM and agentic workflows. Factual · 2 Jun 2026 AI capabilities
Not researched Factual 2 Jun 2026 Microsoft is working on quantum computing with Algorithmic, Columbia and ETH Zurich. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 Microsoft created a new state of matter, theorized a hundred years ago, and proved that it exists. Factual · 2 Jun 2026 Science and research methods
Not researched Factual 2 Jun 2026 Majorana 2 implements a next-generation material stack that Microsoft used Discovery to discover and build and to help fabricate. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 Windows agent containment supports process-level isolation for lightweight agent actions and session-level isolation for user separation. Factual · 2 Jun 2026 AI safety and alignment
Not researched Factual 2 Jun 2026 Microsoft introduced a new GitHub Copilot app on 2 June 2026. Factual · 2 Jun 2026 Their own work or organisation
Mostly true Factual 2 Jun 2026 With the Rayfin SDK, apps built in Replit can be deployed with their data into an enterprise-managed Fabric tenant. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 Foundry has a built-in self-improvement loop so agents continuously get better. Factual · 2 Jun 2026 AI capabilities
Not researched Factual 2 Jun 2026 Outlook Summarize uses on-device AI running locally. Factual · 2 Jun 2026 Their own work or organisation
Not researched Factual 2 Jun 2026 Majorana 2's qubit design makes it possible to fit a million qubits on a chip smaller than a credit card. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 Microsoft and Nvidia built the first supercomputer together to train the GPT models. Factual · 2 Jun 2026 AI compute, hardware and data
Not researched Factual 2 Jun 2026 Majorana 2 qubits maintain their state much longer than other common approaches, which deliver lifetimes of only microseconds or milliseconds. Factual · 2 Jun 2026 Technology
Mostly true Factual 2 Jun 2026 Microsoft announced HorizonDB, a fully managed PostgreSQL service on Azure. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 Web IQ is best in class in quality, speed and cost. Factual · 2 Jun 2026 AI capabilities
Not researched Factual 2 Jun 2026 Azure spans more than 500 data centers in 80 regions. Factual · 2 Jun 2026 AI compute, hardware and data
Not researched Factual 2 Jun 2026 The Rayfin SDK is available now for others to use with their tools as a backend service. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 Microsoft Foundry has the largest model catalog available, including OpenAI, Anthropic and Microsoft MAI models. Factual · 2 Jun 2026 Their own work or organisation
Resolved true Factual 2 Jun 2026 MXC is a new policy layer that lets Windows apply isolation and containment using AI-native or other OS-native primitives. Factual · 2 Jun 2026 AI safety and alignment
Not researched Factual 2 Jun 2026 Microsoft announced two new models that are going to run in box on Windows. Factual · 2 Jun 2026 AI capabilities
Not researched Factual 2 Jun 2026 Benchmarking Cobalt with GitHub Copilot agentic traces shows 33% lower latency, 14% faster speed and 23% higher throughput for agent calls. Factual · 2 Jun 2026 AI compute, hardware and data
Not researched Factual 2 Jun 2026 Microsoft announced WSL Containers on 2 June 2026. Factual · 2 Jun 2026 Their own work or organisation
Not researched Factual 2 Jun 2026 When Microsoft debuted the MDASH harness, it ranked top on the CyberGym benchmark. Factual · 2 Jun 2026 AI safety and alignment
Not researched Factual 2 Jun 2026 Microsoft announced its first quantum processing unit (QPU) in 2025. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 Microsoft is working with many partners so its containment supports real developer workloads. Factual · 2 Jun 2026 AI safety and alignment
Not researched Factual 2 Jun 2026 Microsoft developed the MRC architecture and rebuilt how traffic across Azure moves to support AI workloads. Factual · 2 Jun 2026 AI compute, hardware and data
Mostly true Factual 2 Jun 2026 Developers can now use a full local agentic loop with tool access to build on-device agentic applications without round trips to the cloud. Factual · 2 Jun 2026 AI capabilities
Not researched Factual 2 Jun 2026 Microsoft's new AI accelerator delivers 30% better tokens per dollar than the leading GPU today. Factual · 2 Jun 2026 AI compute, hardware and data
Not researched Factual 2 Jun 2026 Microsoft Discovery combines models, HPC compute, scientific knowledge graphs, and automated lab and simulation into one agentic discovery loop. Factual · 2 Jun 2026 Science and research methods
Not researched Factual 2 Jun 2026 In May 2026 Microsoft announced MDASH, a multi-model agentic security system. Factual · 2 Jun 2026 AI safety and alignment
Not researched Factual 2 Jun 2026 Microsoft introduced Microsoft Execution Containers (MXC) on 2 June 2026. Factual · 2 Jun 2026 AI safety and alignment
Unresolvable Factual 2 Jun 2026 Agents built in Foundry hosted agents have tools, durability, memory and state. Factual · 2 Jun 2026 AI capabilities
Not researched Factual 2 Jun 2026 GPU acceleration in Microsoft Fabric's data warehouse is delivering a 7X performance gain. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 Microsoft added more data center capacity in the last 18 months than in the first decade of Azure. Factual · 2 Jun 2026 AI compute, hardware and data
Not researched Factual 2 Jun 2026 Nvidia is bringing OpenShell to Windows to securely execute autonomous AI agents. Factual · 2 Jun 2026 AI safety and alignment
Own records only Factual 2 Jun 2026 Nadella said that in Microsoft's internal testing, Horizon is delivering something like 3X the throughput of any self-managed PostgreSQL setup. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 Microsoft's Scout autopilot joins group chats in Teams and handles threads in Outlook. Factual · 2 Jun 2026 AI capabilities
Not researched Factual 2 Jun 2026 Microsoft was the first cloud to bring up Nvidia's Vera Rubin system for validation. Factual · 2 Jun 2026 AI compute, hardware and data
Not researched Factual 2 Jun 2026 Microsoft is rolling out vertical taskbars in Windows 11. Factual · 2 Jun 2026 Technology
Not researched Factual 2 Jun 2026 Fairwater was designed from the ground up for AI, in close work with Nvidia. Factual · 2 Jun 2026 AI compute, hardware and data
Not researched Factual 2 Jun 2026 Microsoft is validating Nvidia's Vera Rubin systems as of June 2026. Factual · 2 Jun 2026 AI compute, hardware and data
Not researched Factual 2 Jun 2026 One of the new Windows in-box models is Aion Instruct, a new small language model for reasoning. Factual · 2 Jun 2026 AI capabilities
Back to the list

Factual AI capabilities

“I was able to use a model that is using 500B, I was able to use a 5B, and have it really deliver the same outcome, why would I not use that?”

Satya Nadella, Microsoft CEO 4 June 2026 An Interview with Microsoft CEO Satya Nadella About Finding Core Competencies (Stratechery)

In a Land O'Lakes agent example shown at Build, a 5B-parameter model delivered the same outcome as a 500B-parameter model.

Mixed Partly true and partly false in roughly equal measure.

The second judge would not resolve this claim; the verdict shown is the first judge's.

Said for an organisation they lead or work for; it counts on both records.

What would count as true or false

Judged as of 2026-06-04. True: Microsoft Build 2026 materials or coverage show a Land O'Lakes agent example in which a model of about 5B parameters delivered the same, or equivalent, outcome as a model of about 500B parameters. Mostly true: the example exists, but the parameter counts differ somewhat (e.g., a small model of a few billion parameters against a frontier model), or results were 'comparable' rather than identical. Mixed: a Land O'Lakes small-model example was shown, but the comparison with the large model is not documented. Mostly false: the example was hypothetical, or involved a different customer or setup. False: no such example was shown, or it showed the small model performing materially worse.

Evidence

  1. Here's Everything Announced at Microsoft Build 2026 eWeek (The Neuron) · 3 June 2026
    “The demo then used Land O’Lakes as the example. The presenter showed how an environment can include skills, knowledge, and tools.”
  2. Here's Everything Announced at Microsoft Build 2026 eWeek (The Neuron) · 3 June 2026
    “The presenter said almost 80% accuracy is not good enough for those tasks, so Microsoft is extending the definition of skills to include rubrics on “what good looks like.””
  3. Frontier Tuning: Microsoft Build 2026 daily.dev (Microsoft Developer source) · 3 June 2026 Own record
    “A demo with Land O'Lakes showed the tuned MAI model achieving over 90% accuracy on complex reporting tasks, estimated at 10x efficiency over baseline models.”
  4. What’s 🔥 in Enterprise IT/VC #501 What's Hot in Enterprise IT/VC (newsletter) · 6 June 2026
    “Microsoft claims that Land-O-Lakes has a 10x more cost efficient model than Open AI’s GPT 5.5.”

Reasoning

Coverage confirms a Land O'Lakes agent example was shown at Build 2026 (E1, E2). A Microsoft Developer summary says a tuned MAI model reached over 90% accuracy on the reporting task at an estimated 10x efficiency over baseline models (E3). A newsletter reports Microsoft's claim that the model was 10x more cost efficient than GPT 5.5 (E4). No source gives about 5B or 500B parameter counts or shows the two models reaching the same outcome, so the specific comparison in the claim is not documented.

Cite this claim

Satya Nadella, “In a Land O'Lakes agent example shown at Build, a 5B-parameter model delivered the same outcome as a 500B-parameter model.” (said 4 June 2026). Verdict: mixed. Reliability API, claim c_11bd759401ec, method 0.1.0+28da144b. https://epistacklabs.com/claims/c_11bd759401ec

Claims without a verdict

Is a verdict here wrong, or do you speak for Satya Nadella? Reply on the claim's own page, with the form at its foot (how replies work). Verdicts are made automatically; how verdicts and scores are made.