Amazon finalized a $50 billion investment in OpenAI as AI-chip usage doubles every nine months. Two containment escapes at Anthropic and OpenAI, plus FBI-confirmed cyber-attacks on U.S. water utilities across seven states, raise safety concerns. Google is shifting from frontier research toward robotics and world-model simulations.
Chapters
And welcome to What the Fudge for August 2nd, 2026. This is the Do You Know Where Your Models Went Today edition. So this week we're going to cover a couple things in the areas of all things AI infrastructure, the buildout that's taking place, the security and safety incidents that have been occurring in the news this week, as well as looking at some of the shifting financial realities for what's going on with large tech companies and some of the funds that were of interest in prior updates I provided. Again, we're going to try to keep this under 5 minutes, but starting off, we're looking at Amazon finalizing $50 billion in investment for OpenAI, I think takes it to a 5% position. The analysts from Epic AI have concluded that AI chips in use are going to double every 9 months. So trying to kind of create a next version of a Moore's Law, perhaps. Oracle's doing debt field strategy to get their OpenAI Project Stargate things, I think, kicked into a higher gear. But if you look at what's going on globally, it looks like there's a bifurcation between how the US is doing closed systems around AI and Chinese delegations to a recent UN summit were talking about taking the, um, open source AI approach. So the developing world will probably make the determination of whether closed or open will be the end result in the future, or maybe there'll just be a balance. I, I'm not really big on the whole It's this or that binary, you know, black, white. It's never that. It's always gray. So when I think of the implications for security and safety, two things come into focus are now we've got both Anthropic and OpenAI disclosing their agents escaped containment, some kind of a cybersecurity exercise, and apparently decided to attack external organizations. You know, it happens. And so one question here is just the, the, the punditry is starting to ramp up around whether the circular funding of this entire ecosystem is questionable, whether there's systemic risks to keep in mind, relying on, relying on AI to actually self-monitor. Is that a good idea? And so with that in mind, it's also important to understand that it's not just theoretical in a, in a lab or Frontier Lab. It's actually happening. I believe the FBI has now confirmed multiple physical infrastructure attacks on key infrastructure, specifically US water, wastewater utilities. At least 7 states are reporting cyberattacks, and allegedly there is an operational disruption, or if you think about water flooding. So big question there is, is some of this the cat out the bag, toothpaste out of the tube, pick your metaphor, but I don't know that AI is going to clean up water. When it floods. So something to think about as we go into the world of cyber to now kinetic implications. And then last, a couple things here just on the markets and the volatility associated with what's going on with all these AI capital expenditures. Amazon reported really increased growth in AWS, and probably not due to NAT gateways alone. If you're familiar with the, uh, duckbill group and some of the humor around the, uh, inability to parse an AWS bill. It's so complex, but definitely growing. And think of that as kind of like a wow, it's, it's still moving. And on the other hand, you may remember I mentioned Situational Awareness, which was a hedge fund, they actually recorded a loss in July. They're still up for the year apparently, but a lot of their reasons for the loss were either speculative AI investments or just frankly being new to the whole hedge fund space.. And then if you look at Google, they're now diverging away from Frontier Lab for the sake of Frontier Lab and appear to be going after the real world, meaning like world models to simulate what's going on in physical reality and use that for robotics and the, again, applying the what's now in a Frontier Lab now to the kinetic edge. And then the last thing I just want to wrap up here, keep it under 5 minutes, is EU is going to be taking the DSA or Digital Services Act. Those rules are going to start applying more widely. I think also if you look at even some of the environmental regulations being waived here in the United States, there's still going to be challenges associated with the cost of permitting for anything related to AI infrastructure, specifically power buildouts, you know, so-called DUST permits used to determine where data centers are being built, yes or no. And then lastly, just this whole geopolitical dynamic around, you know, the US and China distillation, their ability to make chips. So I think that's going to kind of increase. But again, trying to keep this under 5 minutes, I did a pretty good job this time. So thank you for listening to What the Fudge for August 2nd, 2026. Be safe, be well, take care. Bye. Bye-bye.
