Microsoft weighs fewer neocloud partners for AI capacity
Microsoft is evaluating a smaller neocloud roster while expanding its own datacentres. The decision affects GPU capacity planning for Azure and AI workloads.
Largest Microsoft Business and Technology Event and Community in Europe • 6.8K Members
Microsoft is evaluating a smaller neocloud roster while expanding its own datacentres. The decision affects GPU capacity planning for Azure and AI workloads.
Executives say AI is increasing the speed of cyberattacks, while defenders use it for monitoring, patching and recovery.
OpenAI is reported to be using NDAs around data centre work. For enterprise teams, the issue is visibility into the infrastructure behind AI services.
Microsoft says threat actors used ARTEX to find public GitHub secrets, then abused application credentials to reach Azure, Microsoft 365 and Azure AI resources.
Microsoft has made the Foundry extension bundle for Azure Developer CLI generally available, adding hosted-agent setup, local testing and deployment through azd.
Microsoft says Foundry Local can run Qwen3.8-27B for GitHub Copilot coding on NVIDIA GPUs with at least 24 GB of VRAM.
October updates add a Copilot harness preview, GA browser automation and hosted-agent resilience, plus Decisions API changes and SDK migration work.
Microsoft has rejected an option to expand a Texas AI compute rental deal with Nscale, adding pressure before the UK start-up’s US IPO.
Apple reportedly reduced some engineering AI allowances from USD 300 to USD 200 a day and is weighing lower caps for other teams.
Installable skills give compatible coding agents Microsoft-maintained guidance for Agent Framework projects in .NET and Python. They are development-time aids, not runtime app skills.
Trump’s super-intelligence loyalty test brings political allegiance into AI policy, raising governance questions for teams working on advanced AI systems.
A Fabric preview creates Real-Time Dashboards from natural-language prompts against Eventhouse or KQL Queryset data, subject to admin and capacity settings.
What does it take to build a conference programme that earns its highest rating yet?
The term moves agentic AI from chat into web work. Enterprise teams should treat it as an operating model question before looking for a product release.
Santander, Novartis and Standard Chartered cited AI gains in third-quarter results. The figures remain uneven, but scrutiny is moving to revenue, costs and margins.
OpenAI’s newest round values it at USD 1.25 trillion, below earlier expectations, as investors weigh AI spending and public listing timing.
AI systems are being used to improve AI research and development. The main signal is a chart tracking longer software tasks and claimed gains in task length.
Hyperscalers plan more than USD 740 billion in capital expenditure this year, while investors focus on free cash flow, debt and compute capacity.
Alphabet's earnings revived questions about AI capital spending, while Microsoft showed stronger cloud growth. The market reaction matters for cloud capacity and AI budgets.
Microsoft’s 29 October assessment says North Korean actors use generative and agentic AI for reconnaissance, lures, tooling and operational tasks.