Cloud Wars Live with Bob Evans

00:03 — According to recent reports, Microsoft is set to reveal its Maia 300 AI accelerator as early as September this year. For context, Microsoft launched Maia 200 back in January with the direct aim of reducing the cost of AI inference.

00:27 — The Maia 300 is expected to support both AI training and inference, but the key story here is the scale of ambition behind it. Now, Microsoft is once again partnering with TSMC, Taiwan Semiconductor Manufacturing Company, who delivered the first iteration of Maia chips. However, this time, the scale is much larger. increasing from tens of thousands of Maia 200 accelerator units.

00:54 — It's increasing from tens of thousands of Maia 200 accelerator units to potentially hundreds of thousands, or even over a million Maia 300 chips. For me, this initiative is really about Azure. By investing in these chips, Microsoft isn't necessarily trying to compete with NVIDIA. Rather, it aims to ensure that Azure doesn't rely entirely on NVIDIA infrastructure.

01:14 — This move will give Microsoft greater control over the cost, supply, and performance of the AI infrastructure running on Azure. And beyond that, Microsoft CEO Satya Nadella has stated that building the company's own silicon to work alongside NVIDIA and AMD chips is a priority.

01:35 — The successful launch of Maia 200 demonstrated that Microsoft's custom silicon strategy can work, albeit on a relatively small scale. The hope now is that Maia 300 will enable Microsoft to scale this strategy into a genuine competitive advantage for Azure.

Visit Cloud Wars for more.

Visit Cloud Wars for more.

Bob Evans speaks with Chad Wahlquist, an Architect at Palantir, about what is driving the company’s extraordinary growth and, more importantly, what customers are getting from its technology. Wahlquist argues that Palantir’s momentum comes from helping enterprises solve difficult operational problems rather than simply deploying AI or chasing model benchmarks. Their conversation explores AI sovereignty, business outcomes, Palantir’s Ontology, LLM complexity, customer operating leverage, and the importance of retaining control over enterprise decision-making.

Outcomes Over AI Hype

 

The Big Themes:

Outcomes Drive Palantir’s Growth: Wahlquist connects Palantir’s growth to the tangible returns customers see after adopting its technology. Rather than treating AI as a standalone investment or another piece of enterprise software, customers increasingly expand their Palantir relationships because successful initial projects create opportunities for broader deployment. He points to strong net dollar retention as evidence that existing customers are spending more after experiencing ROI. The underlying philosophy is straightforward: when an investment generates meaningful business value, executives are willing to repeat and expand it.

Production LLMs Create New Problems: Getting an LLM working is only the beginning. Wahlquist discusses the stochastic and probabilistic nature of models, changing provider guardrails, security requirements, model deprecations, and unpredictable behavior across edge cases. Technically switching from one model to another might appear simple, but ensuring that a replacement works reliably across production workflows is significantly harder. This creates a fundamental enterprise question: how do businesses build durable operations on technology whose behavior and availability can change?

Protect the Enterprise Decision Loop: One of the conversation’s most important ideas is that a business can be understood as a collection of decisions. Companies continually observe data, apply logic, take actions, measure outcomes, and adjust future decisions. Wahlquist calls that feedback loop a source of business “alpha” — the proprietary knowledge that helps one organization outperform another. As AI agents become participants in enterprise decision-making, ownership of that loop becomes increasingly important. Companies should understand who controls the data, logic, actions, outcomes, and learning generated through those processes.

The Big Quote: “LLMs don’t just magically fix everything. They also create new problems that you have to go solve.”

Visit Cloud Wars for more.

00:03 — We've had a lot of big numbers come out for Q2 reports. Google Cloud is at 82 percent. AWS growth jumped up to 37 percent. But I have to take a little more time here to dig into what I'm calling the Palantir phenomenon. Here's a company that, over the last three quarters, has seen its growth rate go from 71 percent to 85 percent to 93 percent.

01:12 — What are customers asking Palantir right now to do for them? That may seem like a goofy, simple question. But if you grow at 93%, you have to be doing something special. What is it that customers are asking for, and that Palantir is delivering, that drove this 93% growth? Chad goes into a lot of detail about that and offers some unique insights there.

02:34 — I think that that's still valid, but the bigger issue now around sovereignty is who controls your data, and not only the data, but how that data is being used to drive AI output. Palantir is making the case that the large language model companies are incorporating the data and everything that goes with it into their own models.

03:27 — Finally, we had a fun conversation considering this company that has grown 93%. It is predicting that the U.S. commercial business, for the next year, will grow 134 percent, but Palantir doesn't have prices for its software.

04:00 — They go in. They talk about what the customer wants to do. They say, "Okay, given that, if we drive the value you're looking for, you pay us this." And then, as that seems to be working quite well, the customer often comes back and says, "Hey, that was terrific. Let's do this, this, and this in addition."

Visit Cloud Wars for more.

00:03 — In a big coup for Salesforce, the Pentagon has approved the use of its Missionforce national security platform to launch AI agents on the most sensitive unclassified missions, and that's because it's been granted Impact Level Five, or IL5 , authorization.

00:21 — The first major user will be Army Human Resources Command, which will use the technology to help with soldier and family support and everyday admin tasks. Later, there will be other use cases, including recruiting, personnel management, and logistics. The big thing here is that Salesforce AI can now take actions and automate workflows.

00:47 — Handling routine tasks while human specialists can stay in the background and remain responsible for sensitive decisions. For the Pentagon, this is quite a jump, moving from AI that, beforehand, solely advised to AI that can actually execute tasks autonomously in military operations. That's no small thing, and I think the fact that Salesforce has won this contract is a massive reflection.

01:14 — On the level of security and governance that the company has in place around its agentic infrastructure, and that will certainly be an attractive proposition for customers. Government contracts always help to boost the credentials of a company, but here, in the age of AI, to have a government department okay the use of agents in an autonomous capacity is a very big win indeed.

Visit Cloud Wars for more.