The Age of Execution: Today’s Most Insightful Tech & AI Twitter Chatter

Supabase just open-sourced a benchmark that runs AI coding agents against its own real-world tasks, and publishes the scores for everyone to see. That move connects directly to what we talked about in episode 143. Last week, it was the EU AI Act forcing an age of consequence with new laws. This week, you're seeing the consequence play out in the market—a demand for PROOF, not just promises. The age of hype is over. The age of execution has begun. Here’s the sweep of what’s moving today. First, the money is following the metal. Three separate funding rounds for robotics and physical AI just crossed a combined one-point-nine BILLION dollars. The key phrase you're hearing isn't "valuation." It's "deployment." The question is no longer how much a company is worth on paper, but how many robots it can actually get into the field.

Companies like Unitree are reportedly making real progress with industrial customers, turning prototypes into products. This isn't science fiction. This is stackable, deployable, physical AI hitting the factory floor. Next, the entire conversation around AI cost is changing. For the last few years, it's been all about the GPU. Who has the most H100s. Who has the biggest training cluster. That's over. The new metric, the phrase you need to know, is "tokens per watt per dollar." It's a measure of INFERENCE efficiency. How much useful work can you get out of your hardware for every dollar you spend and every watt you burn? This isn't just about Nvidia anymore. It's about the entire physical infrastructure chain.

Behind every chatbot response sits a stack of computing power, high-speed networking, advanced cooling systems, and data-center equipment that all need to work in perfect concert. The focus is shifting from the star performer—the GPU—to the entire orchestra. And that brings us to the sleeper conversation gaining momentum. It’s about chip design. Specifically, AMD. There's a growing, detailed argument making the rounds that AMD's chiplet architecture—their modular design—is perfectly suited for where AI is going next. The argument is that this approach gives them a fundamental advantage in cost, manufacturing yield, and flexibility. It’s a technical argument, but the implication is huge.

As AI moves from a few massive training runs to trillions of daily inference requests, the economics of producing chips at scale becomes the dominant factor. The discussion is shifting from raw power to economic reality. And finally, there's the benchmark that started us off. Supabase Evals. It’s not just another leaderboard. By open-sourcing a test that runs agents like Claude Code and Codex against REAL, messy, everyday programming tasks, Supabase is sending a clear message. The lab is closed. The real world is the only test that matters now. So let's dive deeper into the two big shifts this week: the new economics of AI, and the relentless push into the physical world. They are deeply connected.

Let's start with the money. That one-point-nine billion dollars flowing into physical AI isn't just another venture capital headline. It's a signal. For years, AI progress was measured by parameters and benchmarks on abstract tasks. Now, the value is shifting to deployment. To things that move, that build, that interact with the messy, unpredictable physical world. This is a much harder problem. And it changes where the innovation needs to happen. It's not just about a better model, it's about better sensors, more durable hardware, more efficient motors, and software that can handle the chaos of reality. When you see a company like Unitree, which makes quadruped robots, getting traction with industrial clients… that’s not a demo.

That’s a manufacturing line. That’s a logistics warehouse. That’s AI with a job to do. And this push into the real world is what’s forcing the second, even bigger shift. The radical change in how we think about the cost of AI. For the past few years, the entire industry has been obsessed with training. The massive, power-hungry, eye-wateringly expensive process of creating a foundational model. Inference—the act of actually USING the model to answer a question or generate code—was seen as the cheap part. The after-thought. That assumption is now being completely dismantled. The new reality is something called Agentic AI. Autonomous agents. Think of fleets of AI workers that you give a task to, and they work on it 24/7.

They don't just answer one query. They orchestrate tools, they reason, they move data, they run multiple processes in parallel to achieve a complex goal. And here's the thing. This kind of work… it doesn't just need GPUs. It needs a TON of CPU power for coordination, for running what are being called "agent sandboxes." It's a completely different kind of workload. One analyst on Twitter, MikeLongTerm, put out a detailed thread on this, projecting that agentic AI will push inference from a small fraction of AI compute to over ninety percent within two years. Ninety. Percent. So what does this all add up to? It means the entire economic model of AI is flipping on its head. The focus is no longer on the one-time cost of training a model.

It's on the continuous, operational cost of running millions of agents, performing trillions of tasks, every single second of every single day. And when you're operating at that scale, tiny efficiencies matter. A one percent improvement in cost isn't a small win. It's a game-changing competitive advantage. This is why you're suddenly hearing about "tokens per watt per dollar." It's a ruthless measure of efficiency. It's why the conversation has expanded beyond the GPU to the entire data center stack—the networking, the cooling, the power delivery. Every single component in that chain is now a potential bottleneck, or a potential source of competitive edge. It’s a return to first-principles engineering.

Now. This is where the argument around AMD gets really specific. For years, AMD’s CEO Lisa Su has been talking about this shift. She predicted that the opportunity would expand far beyond training and into inference, into enterprise use, into these agentic fleets. And she bet the company's architecture on it. The argument you're seeing online now is that AMD's chiplet strategy is the payoff to that bet. Instead of designing one giant, monolithic processor—which is expensive and difficult to manufacture—AMD builds its processors from smaller, modular "chiplets" connected together. Here’s why that matters NOW. First, cost and yield. Smaller dies have dramatically better manufacturing yields.

When a defect happens on a silicon wafer, it only kills one small, relatively cheap chiplet, not a massive, expensive processor. You get less wasted silicon. You can turn partially defective chiplets into lower-cost products instead of throwing them away. AMD has said publicly that for high core-count chips, this design can cost roughly HALF of an equivalent monolithic chip. In a world where you need to deploy millions of CPUs to manage your AI agents, a fifty percent cost reduction is not a feature. It's a revolution. Second, scalability and flexibility. AMD uses the same core chiplet designs across its entire product stack, from consumer desktops to the most powerful server processors.

This dramatically cuts down on design and manufacturing costs. It allows them to scale up core counts for massive server chips almost linearly, while competitors using older designs see costs explode exponentially. This is the key. The agentic AI future requires a different balance of compute. The old ratio of GPUs to CPUs in a data center is breaking. The forecast is moving toward a one-to-one, or even a one-to-two ratio of GPUs to CPUs. That means the demand for CPUs isn't being replaced by GPUs—it's being ADDED to by this new layer of agentic orchestration. You're looking at a future where monthly token consumption is measured in the quadrillions. One estimate I saw pegged it at 35 quadrillion tokens a month.

That's a 160-times growth in just two years. That level of demand breaks old models of production. It requires a new way of thinking about building the engines of intelligence. The argument is that AMD saw this coming, and built the architecture for THIS world, not the last one. This isn't about fanboy-ism. This is about economics. It’s a structural argument about manufacturing, cost, and scale. And it connects everything we're seeing this week. The push for real-world deployment in robotics, the need for verifiable performance shown by Supabase Evals, and the relentless focus on the cost-per-token... they all point to the same conclusion. AI is industrializing. The period of blue-sky research and dazzling demos is giving way to a new era defined by engineering discipline, supply chain management, and brutal economic competition.

The winners won't be the ones with the flashiest PowerPoints. They will be the ones who can execute at scale, reliably, and more efficiently than anyone else. This week wasn't about a new model launch. It was about the sound of the factory floor getting louder. It was about the shift from what AI could do, to what it actually costs to do it, a billion times a day. The questions have changed. And the companies that have the right answers are the only ones that will matter.