Microsoft: This is how you control AI

Welcome to Runtime! Today: Microsoft strikes a big deal with Anthropic and makes a bid to be the central hub for AI agents, Cloudflare takes out roughly half the internet, and the latest funding rounds in enterprise tech.

(Please forward this email to a friend or colleague! And if it was forwarded to you, sign up here to get Runtime each week.)

Business time

SAN FRANCISCO — As we approach the three-year anniversary of the launch of OpenAI's ChatGPT, which made generative AI pretty much the only thing on the enterprise tech agenda, we still don't know exactly how businesses want to consume these services. The motivation to improve business productivity through agentic AI is clearly there, but it's anyone's guess as to how those agents should be built, managed, and controlled.

Microsoft laid down a bet Tuesday at its Ignite 2025 conference that so long as it provides the right amount of AI options to the corporate world, those businesses will put their eggs in its basket. It introduced a new central hub for managing AI agents, and perhaps most notably announced that it has teamed up with Anthropic to provide its Claude AI models to Azure customers.

• "This partnership will make Claude the only frontier model available on all three of the world’s most prominent cloud services," Microsoft said in a blog post.
• Microsoft said it would invest "up to" $5 billion in Anthropic, which has received considerable investment from AWS and Google Cloud, and Anthropic has agreed to spend $30 billion on infrastructure services from Azure.
• Such a move felt inevitable after Microsoft and OpenAI entered into their polycule era by modifying their partnership agreement, and Runtime just called up our bookie and put a substantial amount of money on OpenAI CEO Sam Altman making an appearance at AWS re:Invent in two weeks to finally bring its flagship AI models to the cloud leader's customers.

After a year of promises and demos, businesses are still trying to figure out whether the hassle of investing in AI agents will ever pay off. It usually takes some time for the tooling to catch up to the hype, and Microsoft unveiled a new service Tuesday called Agent 365 that it thinks will help customers manage agentic AI.

• "Agent 365 delivers unified observability across your entire agent fleet through telemetry, dashboards, and alerts," Microsoft's Charles Lamanna said in a blog post, describing the new service as "the control plane for AI agents."
• The service will allow Microsoft customers to manage both homegrown agents as well as third-party agents developed through partners like Databricks, SAP, and ServiceNow.
• As we've been harping on for a while, and as The Information handily summarized in a report last week, pretty much every company in enterprise software is rolling out agents in hopes that customers will choose their platform as the central hub for those agents.
• "The clearest path forward is to manage agents the way you manage people, using the same infrastructure, apps, and protections that power your business today," Lamanna said.

But that sentiment is far from clear; are agents automated workers, or are they a new type of software tool?  Microsoft's Judson Althoff, who has taken on a much larger role at the company in recent months, argued that The Business — which is how tech people describe non-tech corporate activity — is now in charge.

• "The more you study it, the difference between successful [AI] projects and those that fail [is] this mindset shift from technology-focused efforts to business-led transformation empowered by AI, and in that order," he said.
• Agents have yet to be deployed widely enough across enterprise tech for practitioners to settle on best practices for building, managing, and updating those agents, but there are clearly large portions of the managerial class that believe AI will allow them to get more control over how their companies use technology.
• "Customers seek a place to manage their agents and often start with the vendor with the largest footprint in their tech landscape," Constellation Research's Ray Wang said in response to the news out of Ignite.
• That's good news for Microsoft, but will be interesting to follow given that tech transformations are historically driven by newcomers disrupting the old guard.

Editor's note: Microsoft provided travel and accommodations for Runtime to attend the Ignite conference.

If you value independent enterprise tech journalism, please consider becoming a monthly supporter of Runtime. For $10 a month, you'll help us continue our mission to bring reliable and actionable coverage of this vital sector of the economy and gain access to supporter-only features currently in the works, such as an exclusive discussion and events forum.

Whoops

A widespread outage at Cloudflare Tuesday morning took down a big chunk of the internet, including OpenAI, X (maybe a good thing), and, based on my experience, the network operations of the Marriott hotel chain. The ripple effects of the outage lasted around eight hours until Cloudflare declared that its services were operating normally around 1130am PT.

It took the company well over an hour to identify the issue, which was eventually blamed on "a latent bug in a service underpinning our bot mitigation capability [that] started to crash after a routine configuration change we made," Cloudflare CTO Dane Knecht said on LinkedIn. The outage was not the result of any kind of attack on Cloudflare's network, but "that issue, impact it caused, and time to resolution is unacceptable," he said.

It hasn't been a great couple of weeks for cloud infrastructure services companies after massive outages at AWS and Microsoft similarly affected broad swaths of the internet. Cloudflare is known for its comprehensive outage reports following any disruption, and we'll await further details on what led to the incident. (Updated like two minutes after this published: Cloudflare released the detailed report.)

Enterprise funding

Lambda raised over $1.5 billion (!) in Series E funding, which will help the neocloud keep up with the pace of investment in AI infrastructure for at least a little while.

d-Matrix landed $275 million in Series C funding, which values the AI inferencing neocloud at $2 billion.

Alembic scored $145 million in Series B funding for its "causal AI" technology, which it says helps companies understand how their marketing investments lead to business outcomes.

Sakana AI raised $135 million in Series B funding as it continues to develop a more efficient approach to training AI models.

Span landed $25 million in Series A funding to automate the software development life cycle, focusing on developer productivity.

Runlayer launched with $11 million in seed funding to build a tool that helps enterprises manage and secure MCP servers.

The Runtime roundup

CISA plans to hire more people over the coming year after Trump administration job cuts left it with "an approximately 40% vacancy rate across key mission areas," according to an internal memo seen by Cybersecurity Dive.

Arm will build Nvidia's NVLink technology directly into the cores it designs for its chip-making customers, allowing a wide variety of partners to improve performance when connecting to Nvidia's GPUs.

Fidelity sued Broadcom after it forced the financial management company to adopt one of its higher-priced VMware bundles rather than renewing its existing subscription deal, according to Reuters.

And in other Broadcom news, apparently VMware overestimated the computing resources needed to run vSAN for years, which is welcome news for new customers but perhaps a lawsuit waiting to happen on behalf of customers that spent way more than they needed to.

Thanks for reading — see you Thursday!