Cloud Wars Live with Bob Evans

In today's Cloud Wars Minute, I explain why agent sprawl may become one of the biggest hidden risks of the AI Era.
Highlights
00:03 — The massive increase in the adoption of agentic AI technology will have significant consequences for businesses. There’ll be increased productivity, the ability to reskill employees who have more time to focus on other areas of the business, and opportunities to explore new business ideas and avenues.
00:32 — And something else — a surge in the number of AI agents, millions, millions, and millions of them. All of these agents lead to a phenomenon called agent sprawl. So if data sprawl is diesel-driven, think of agent sprawl as running on jet fuel. Without proper governance and visibility, this can lead to shadow AI, which, in the wrong hands, could effectively bring down a business.
01:04 — To avoid this, Salesforce has added automated discovery for AI agents and tools to MuleSoft Agent Fabric. Andrew Comstock, SVP and GM of MuleSoft at Salesforce, said "The expanded capabilities give you the freedom to innovate across any platform while maintaining the unified visibility and control needed to scale."
01:32 — At the core of these enhancements are agent scanners, which automatically detect and catalog AI agents across Salesforce Agentforce, Amazon Bedrock, Google Cloud’s Vertex AI, and other authorized AI platforms. Additionally, for MCP services and other bespoke agents, MuleSoft Agent Fabric facilitates easy integration across a company’s entire agent ecosystem.
02:01 — The recent updates replace manual oversight with automation, enhance security by providing instant visibility into multi-cloud agents, highlight internal tools that may otherwise be overlooked, and offer a unified agent map to identify areas for optimizing AI investments.

Visit Cloud Wars for more.

Highlights
00:02 — A month ago, I moved Google Cloud up to the #1 spot on the Cloud Wars Top 10, moved Microsoft down to #3. That was predicated in large part on the tremendous job Google Cloud has done in building sort of the twin pillars: AI and cloud, the way its customers are building for the future, not just perfecting what they've done in the past.
00:27 — And also the company's impressive growth rates, showing that more and more, in spite of the size differential between Microsoft and Google Cloud, Google Cloud was winning a disproportionate share of new business, showing it is becoming, back then, the favored cloud and AI vendor.
00:47 — Well, the Q4 numbers for Google Cloud came out yesterday, and there's no doubt that that was the right call To make. Google Cloud's Q4 revenue jumped 48% to $17.7 billion and they beat Microsoft for the first time ever in a very key metric, it's the really the big thing I want to talk about here today.
01:50 — Microsoft's last three quarters, it grew 27% 26, 26. For Google Cloud, it's 32, 34, 48. Google Cloud is on a massive acceleration run here. So, in spite of the fact that the numbers here, the revenue figures, are different, what we see is Google Cloud accelerating wildly. Well, Microsoft has leveled off again.
02:23 — The key point here, this key metric I talked about up above, if you look at the incremental revenue gains each company made, looking at what calendar Q3 ended, September 30 to calendar Q4 December 31, those are the periods we're looking at. Google Cloud's revenue Q3 to Q4 went up $2.5 billion from just over $15 to $17.7, Microsoft's went up $2.4 billion $49.1 to $51.5.
03:30 — While the heart of this discussion is around Google Cloud, Microsoft has been doing an extraordinary job in this very competitive market, but that's why I call it the greatest growth market the world has ever known. We're seeing companies perform in this market unlike any other industry at any time in human history.
04:47 — The big thing about it here is you've got these smaller, disruptive cloud and AI players, Google Cloud at #1, Oracle #2, Microsoft down at #3. I moved AWS down to #7. It's doing some really good things in a lot of ways, but as far as the company setting the agenda in line with their customers for the future for the AI economy, it's Google #1, Oracle #2, Microsoft #3.

Visit Cloud Wars for more.

In this episode of the AI Agent & Copilot Podcast, John Siefert, CEO of Dynamic Communities and Cloud Wars, sits down with Dona Sarkar, Chief Troublemaker, Enterprise AI Advocacy at Microsoft, to explore what it really takes to move AI agents and copilots from experimentation into production. Their conversation previews Sarkar’s keynote at the 2026 AI Agent & Copilot Summit NA and dives into practical adoption, human-centered AI, and lessons learned from real-world enterprise deployments.
Key Takeaways
Enterprise advocacy bridges the gap: Sarkar explains that enterprise cloud advocacy exists to translate Microsoft product capabilities into practical, real-world business solutions. Rather than selling tools, her team focuses on enablement — creating demos, workshops, and labs that show how AI agents, Copilot Studio, Azure, and Power Platform can actually be deployed inside organizations.
Production is harder than experimentation: Building an AI agent is easy; deploying it responsibly is not. Enterprises struggle with permissions, ownership, data readiness, and governance once agents move into production. These challenges reveal why successful AI adoption requires cross-functional collaboration between IT, business units, and governance teams.
Not all work should be automated: Sarkar cautions against replacing meaningful human interactions with automation simply because it’s possible. Instead, organizations should focus AI on prioritization, analysis, and repetitive tasks — freeing humans to spend more time on creativity, judgment, and relationship-building. “We really need to go draw a big old line in the sand and say, these should be uniquely human to human activities," she says. "These should be uniquely AI to human activities. These should be uniquely AI to AI activities.”
Human connection matters more than ever: Despite fears that AI would reduce in-person interaction, both speakers observe the opposite trend. Conferences and professional gatherings are thriving because people crave perspective, not just information. While AI can surface data instantly, point of view comes from lived experience.
Failure is part of responsible AI adoption: Sarkar openly shares that "The number of agents I’ve had to take down is probably like 50% of the agents I built.” These failures weren’t wasted effort; they informed better tooling, clearer governance, and improved workflows. Microsoft’s rapid release of new AI tools reflects lessons learned internally before being shared with customers.

Visit Cloud Wars for more.

In today’s Cloud Wars Minute, I look at how Microsoft is helping developers build and scale AI agents safely inside Visual Studio Code.
Highlights
00:10 — The Microsoft Copilot Studio extension for Visual Studio Code is now generally available, providing developers with the ability to build and manage Copilot Studio agents directly within the IDE. This extension is designed for developers and integrates seamlessly into their workflows.
00:28 — It includes standard Git integration, request-based pull reviews, auditability, and is tailored to the VS Code UX. The new extension reflects the growing complexity of agents and equips developers with the same best practices they use for app development, including, as Microsoft puts it, source control, pull requests, change history, and repeatable deployments.
01:02 — This extension really benefits developers when they need to manage complex agents, collaborate with multiple stakeholders, and ensure that any changes made are done so safely. It’s ideal for developers who prefer to build within their IDE while also having an AI assistant available to help them iterate more quickly and productively.
01:30 — The extension introduces important structural support for the development of AI agents. By integrating Copilot Studio directly into VS Code, Microsoft is empowering developers to build more efficiently, without compromising control, access to collaborators, or safety. This is a critical combination as AI agents become increasingly more powerful and complex.
02:00 — As these agents continue to evolve, they require the same stringent checks and balances as traditional software. Microsoft’s Copilot Studio extension addresses this by giving developers the tools they need to scale agents responsibly while maintaining performance.

Visit Cloud Wars for more.

In today’s Cloud Wars Minute, I break down Palantir’s breathtaking Q4 results and why its customer-first AI approach is a wake-up call for the entire enterprise software industry.
Highlights
00:06 — Several companies in the Cloud Wars Top 10 have been rolling out their numbers for Q4 and for year 2025, the numbers from Palantir, expected to be pretty good. They're absolutely stunning. Total revenue growth for Q4 was up 73% and their guidance for Q1 indicates more of the same. They are guiding to 73% revenue growth for the quarter we're currently in.
00:54 — The numbers are just absolutely extraordinary, I think, breathtaking. There's no talk from Palantir that you hear from so many other enterprise software companies: "Oh, it's a challenging macroeconomic environment. There's global uncertainty, there's budget pressures on." Palantir just doesn't talk about that.
01:39 — Palantir comes in and says, "What are you trying to achieve? What are your business outcomes? Let's start there, and then we'll back up, use some of the existing software we have, or some variations combinations of it, what we're doing and with partners, it is the way of the future."
02:27 — Now let me just step away from Q4 a second and go to the full year. For the full year, revenue is up 56%. So, that means that throughout 2025 the growth rate for Palantir is accelerating significantly. Generally, we would see this flipped, where a company grew very fast in the first few quarters, but understandably slowed down as the revenue basis got bigger.
04:10 — They don't fall into the trap of having to engage with customers based on categories or boxes or buckets of industry terms created by the big analyst firms ... They do not go down that path. They say each customer's problems are unique. The capabilities that we bring are unique. The engagement model is unique, and the outcomes we want to work with are different here.
04:49 — So I think that behind these breathtaking numbers there's a wake-up call for all of the big enterprise software companies, right? What got us here will not get us there? And I think Palantir is offering to everybody an example of the new type of thinking technology and engagement models that are going to be required here into the AI economy.

Visit Cloud Wars for more.