Today in AI: Data Centers Surge, Cybercab Rolls, and AGI Nears

Ada: Welcome to Today in AI, I'm Ada.

Sam: And I'm Sam. Today, we're tracking a six-billion-dollar data center driving a land rush, Tesla's new Cybercab hitting the streets, and some bold predictions about AGI.

Ada: Let's kick things off with a massive infrastructure story that's shaping the AI landscape. CoreWeave's six-billion-dollar data center in Lancaster is making headlines, not just for its sheer scale, but for the ripple effect it's having. We're seeing a powered-land rush, with data center land deals jumping a staggering seventy-nine percent in early two thousand twenty-six.

Sam: That's an incredible surge, Ada. It really highlights how critical infrastructure is becoming for AI development. Six billion dollars isn't just a big number, it's a statement about the compute power needed to fuel the next generation of models and applications. This isn't just about building a big shed for servers, it's about securing access to reliable, high-capacity power grids, which are becoming increasingly scarce in prime locations.

Ada: Exactly. And that seventy-nine percent jump in land deals tells us that CoreWeave isn't alone in this race. Other players are clearly anticipating similar needs and are aggressively acquiring property to build out their own AI infrastructure. It's a land grab driven by the insatiable demand for GPUs and the facilities to house them. This is a foundational shift, not just a temporary spike.

Sam: From massive data centers to personal transportation, let's switch gears to something a bit more visible on the streets. Tesla's gold Cybercab fleet has flooded Austin, pitching thirty-thousand-dollar two-seaters at half Uber's price.

Ada: Thirty thousand dollars for a two-seater that's half the price of Uber, that's a bold move, Sam. It's an interesting strategy from Tesla, positioning the Cybercab as a direct competitor to ride-sharing services, but with the added appeal of ownership. The gold color is certainly a statement, ensuring they won't be missed on the Austin streets.

Sam: It certainly is. And the price point is key here. If they can truly deliver a compelling experience at thirty thousand dollars, it could significantly disrupt urban transportation. It's not just about autonomous driving, it's about making personal, on-demand transportation more accessible and potentially more affordable than traditional ride-sharing. The question will be about adoption, infrastructure for charging, and how it integrates into the existing urban fabric.

Ada: Moving on to some big pronouncements about the future of AI itself. Nvidia's CEO has declared that AGI is practically here. This comes alongside news that GPT-6 Astra is bringing full computer control to developers, and Databricks is empowering marketers with AI-driven insights. It feels like we're hitting a new gear.

Sam: Jensen Huang's statement about AGI is always going to turn heads. While we covered some of his thoughts yesterday, the continued emphasis on AGI being 'practically here' really underscores the rapid pace of development. It's a strong signal from one of the most influential figures in AI, and it certainly aligns with what we're seeing with models like GPT-6 Astra.

Ada: That's a good point, Sam. The 'AGI is practically here' narrative, combined with GPT-6 Astra offering full computer control to developers, paints a picture of increasingly capable and autonomous AI systems. This isn't just about better chatbots anymore, it's about agents that can interact with and operate within digital environments with minimal human oversight. That's a huge leap in functionality for developers.

Sam: Indeed. And while the promise of AGI is exciting, it also brings us to our next story: growing fears about AI's impact on employment. Seventy-one percent of Americans are backing unions as AFL-CIO President Liz Shuler warns that AI is already displacing back-office jobs without worker guardrails.

Ada: This is a critical conversation that needs to happen now, not later. The seventy-one percent support for unions is a clear indicator that the public is concerned about job displacement. Liz Shuler's warning about back-office jobs being affected without guardrails highlights a real and present issue. We're not talking about hypothetical future scenarios, but actual job shifts happening today. The question is, what kind of guardrails are needed, and how quickly can they be implemented?

Sam: It's a complex challenge, Ada. On one hand, AI offers immense productivity gains. On the other, the human cost of those gains needs to be addressed. The focus on back-office jobs is particularly telling because those are often process-driven roles that are more susceptible to automation. Unions are clearly stepping up to advocate for workers in this evolving landscape, pushing for policies that ensure a just transition rather than just displacement.

Ada: Speaking of productivity, OpenAI says it has hit its research intern goal, with agents now performing at three point one times human workdays and median use over six hundred dollars per day. That's a significant milestone.

Sam: Three point one times human workdays is a remarkable achievement. It shows that these AI agents are not just assisting, but are truly augmenting and even surpassing human output in specific research tasks. The median use cost of over six hundred dollars per day also gives us a glimpse into the operational expenses of running these advanced AI agents at scale. It's expensive, but if they're delivering that kind of productivity, it's a cost many will likely be willing to bear.

Ada: It certainly highlights the growing efficiency of AI in research environments. This isn't just about automating simple tasks, but about achieving a level of output that significantly accelerates the pace of discovery. It's a powerful indicator of how AI is transforming the very process of innovation.

Sam: And on a related note, OpenAI Chief Scientist Jakub Pachocki warns that reasoning models are accelerating toward recursive self-improvement, and chain-of-thought monitoring is fading. This is a fascinating, and perhaps a bit concerning, development.

Ada: Pachocki's warning about recursive self-improvement is a significant one. If reasoning models are indeed accelerating their ability to improve themselves, and our ability to monitor their 'chain of thought' is fading, it raises important questions about control and interpretability. It's the kind of development that underscores the need for robust safety and alignment research as these models become more autonomous.

Sam: Absolutely. The idea of models improving themselves at an accelerating rate, coupled with reduced transparency into their internal reasoning, is a core concern for many AI safety researchers. It’s a delicate balance between pushing the boundaries of capability and ensuring we maintain a degree of understanding and control over these increasingly powerful systems.

Ada: And finally, a quick update on some specific AI development. Peter Gostev says GPT-6 Astra migrated one hundred fifty thousand lines of legacy code without babysitting, but now needs a remote Linux box.

Sam: Migrating one hundred fifty thousand lines of legacy code without human intervention is incredibly impressive. That's a massive amount of work that typically requires significant developer hours and can be prone to errors. The fact that Astra could do it autonomously speaks volumes about its capabilities. The need for a remote Linux box is a minor operational detail compared to the scale of the code migration achievement.

Ada: It really showcases the practical application of these advanced models in real-world development scenarios. Code migration is a tedious but essential task, and if AI can handle it, that frees up human developers for more complex, creative work.

Sam: That wraps up our top stories for today. A lot of ground covered, from massive infrastructure to autonomous code migration.

Ada: Indeed. For more details on all these stories and to dive deeper into the world of AI and startups, visit startuphub.ai. I'm Ada.

Sam: And I'm Sam. We'll see you tomorrow on Today in AI.