AI Power Moves: Twitter's Smartest Tech Chatter, Curated Daily

A new AI model with two-point-eight trillion parameters just ranked number four in the world. And it doesn't just write text. It builds entire video games. Last week on the show, we talked about the power plays fueling the future of AI. This week, we're seeing what that future actually builds—and it's not just about who has the most chips, but who has the code that can create entire worlds from scratch. This is the new baseline. The conversations on Twitter, the ones that matter, have shifted. It's no longer about whether AI can do a task. It's about what happens when AI can build the entire factory. Here’s the sweep of what’s moving right now. First up, a new job title is quietly emerging.

Tomasz K. Stańczak, a founder from the Ethereum world, is building a new robotics startup at the Hugging Face incubator this summer. The key phrase in his hiring post? He’s looking for “AI-native engineers.” This isn’t just a buzzword. It's a signal. It tells you there's a new kind of developer in demand—one who doesn't just bolt AI onto an existing product, but builds from the ground up with AI as the core foundation. They think in prompts and agents, not just functions and classes. Keep an eye on this. That phrase is going to be on every recruiter's list by next year. Next, the old world continues to create openings for the new. A thread from ABFinance today laid it out plainly: legacy banking models are so inefficient, so hostile to a good user experience, that they are practically begging to be disrupted.

This isn't a new observation, but the tone has changed. It's less about plucky startups chipping away at the edges, and more of a foregone conclusion. The conversation is now about how quickly fintech can absorb the functions of traditional banks, not if it can. Every point of friction in your banking app is a business opportunity for someone else. And this isn't just a Silicon Valley story. Far from it. Take a look at what’s happening with Ignite in Pakistan. This is a public sector company, backed by the Ministry of IT, that's actively funding and promoting tech entrepreneurship. They're running incubators, hosting e-sports championships, and building a real ecosystem. This is critical.

It shows that the playbook for creating a tech hub is being decentralized. You don't need to be in California to build a high-impact venture anymore. The tools, the knowledge, and now, the institutional support, are spreading globally. Innovation is becoming a distributed system. So you have these threads running in parallel. A new kind of AI that builds systems. A new kind of engineer who can wield it. A financial system ripe for overhaul. And new global hubs rising up to compete. So what does it all add up to? Let's go back to that first story. The big one. Kirill, a developer on X, dropped a bomb yesterday. A two-point-eight trillion parameter model. For context, that is an immense, almost unfathomably large model.

It just landed at number four on global independent benchmarks. And the price? A third of its main competitor. But here's the part that you need to focus on. The capabilities. Kirill states it plainly. It builds "full games, 3D data visualizations, and combat engines." Let's break that down, because it's easy to gloss over. This is not an AI that generates a nice 3D model of a character. This is not an AI that writes a piece of dialogue. This is an AI that builds the engine. A combat engine isn't an asset; it's a complex system of logic. It's hitboxes, damage calculation, enemy behavior, physics, and player feedback all working in concert. To say an AI can build one from a prompt is like saying a 3D printer can produce a fully functional Swiss watch.

This is a tectonic shift. For the past few years, we’ve seen generative AI as a tool for content. It makes pictures. It writes marketing copy. It summarizes documents. It was an assistant, a super-powered intern. What Kirill is describing is an AI that is a systems builder. It doesn't just create the assets for the game; it creates the game itself. The underlying logic. The rules of the world. Think about what that means for a software team. The entire workflow is upended. You're no longer meticulously coding every interaction. You're becoming an editor. A director. You guide the AI, you give it the high-level vision, and it generates the complex scaffolding of the application. You prompt a "fun, fast-paced combat engine for a fantasy RPG," and it delivers a working prototype.

Your job then becomes to refine it, to test it, to steer the generation. This collapses the stack. It makes the act of creation exponentially faster and more accessible. But it also raises a terrifying question for anyone whose job is to build that scaffolding manually. When a model can build a combat engine in an afternoon, what is the value of a team of engineers who would have spent six months on the same task? The answer is uncomfortable. Their value shifts from being bricklayers to being architects and quality control. A completely different skillset. And this brings us directly to that second thread. The "AI-native engineer." The phrase is so perfect. Tomasz K. Stańczak is hiring for his robotics startup, and he doesn't just want a software engineer who knows some AI.

He wants someone who is AI-native. So what IS that? Uhm, it's a mindset. It's the difference between someone who grew up with a landline and someone who grew up with a smartphone. The mobile-native generation didn't just think of phones as devices for calls; they saw them as portals to the internet, as cameras, as navigation tools. They built apps like Instagram and Uber that couldn't have even been conceived of in a desktop-first world. The AI-native engineer is the same. They don't see AI as a feature to be added on top of an application. They see it as the fundamental substrate. They assume that uncertainty, probability, and learning are core components of the system, not edge cases to be handled.

They build products that are designed to be steered, not just operated. Why is this showing up in a robotics startup? Because robotics is the ultimate expression of this shift. You cannot write enough if-then statements to make a robot navigate a messy, unpredictable human environment like a kitchen. It's impossible. The only way to solve it is with an AI that can perceive its surroundings, understand goals, and generate its own plan of action in real time. The robot's "software" is a live, running model, not a static block of code. So you need engineers who think that way. People who are comfortable building systems that learn and adapt. People who understand how to design the goals, the rewards, and the constraints for an AI, rather than dictating its every move.

They are less like traditional programmers and more like… trainers. Or teachers. They are building a mind, not just a machine. This is the connection. This is the pattern you need to see. On one hand, you have these monstrously powerful new models like Kirill's, capable of building complex systems on demand. They are the new foundational layer—the engine of creation. On the other hand, you have the emergence of the AI-native engineer, the new kind of builder who knows how to operate this engine. The tools are becoming exponentially more powerful. And the people who can wield them are becoming a new, distinct class of talent. This isn't a future prediction. It's happening right now. The job postings are live.

The models are running. So, where does this leave us? This week's chatter wasn't just noise. It was a signal of a phase change. We're moving from an era of AI experimentation to an era of AI application. And not just application as in "using an API." Application as in building entire new structures, new systems, new realities with these tools. The conversation has matured past the simple awe of a chatbot writing a sonnet. Now, the real question is about building durable, complex, interactive systems. Whether it's a video game, a financial service that bypasses legacy banks, or a robot that can clean your kitchen. All of it requires this new way of thinking. It's a convergence of a new class of tool—the system-building AI—and a new class of builder—the AI-native engineer.

One without the other is just potential. Together, they represent a fundamental change in how we create things in the digital and physical world. The threads are all connected. The fintech disruptors are using AI-native approaches to build more efficient systems. The global tech hubs like the one growing in Pakistan are training a new generation to be AI-native from the start. They aren't held back by legacy thinking. This is the delta. This is what's different today. Yesterday, we were impressed that an AI could draw a picture. Today, we're asking it to build the museum. The debate over AI's potential is over. The race to build with it has just begun.