Congrats @ekuyda very exciting.

If I were a hyperscaler, Nvidia, or another tech company with lots of cash, I would buy Neocloud bonds at 10%+ rates. Strong returns on positive cash balance, and if they go bankrupt, seize their compute and power resources for existing business. Better move than buying equities in these businesses.

I still don’t understand the point of a large institutional investor meeting while Anthropic is in confidential filing. If they wanted a TTW meeting, they should have held it with select institutional investors in their offices. That would have avoided a circus and limited leaks. Given their S-1 leaked to the news media (I’ve never seen another in my 26-year career), the company should have been more sensitive.

The maturity displayed by management teams of public AI and leading software companies versus their private peers is striking.

Companies that control or own compute have a meaningfully strong competitive position versus competitors that do not. wsj.com/tech/ai/ai-computing…

I think Amazon realizes it will massively increase the return rate and will completely mess up their unit economics. Also it will increase fraud risks significantly that many investors are not realizing.

What an amazing couple of days at #Khanference in Dallas. Imran’s (@dottkhan) ability to bring together leaders from business, the military, and government for deep, candid conversations is unmatched. Great people. Great energy.

Great conversation with @jlanzone @Jasonmaynard @niravtolia @1PasswordCEO talked about the rise of consumer agents & what it will actually take for them to work: identity and trust, understanding what we want, personal context, and ultimately human connection Another great Khanference in the books. Thank you @dottkhan!

DFW>LGA … heading home inspired & nourished from @dottkhan #Khanference talks & networking conversations on AI, advertising, agentic commerce, global security, space tech, Ukraine economy and so much more …

Super fun to spend time with oldie but goodie friends @adambain, @joshelman, @jlanzone, and @peterpham during @dottkhan's annual "Khanference" this week in Dallas!

XNRGY founder and CEO Wais Jalali on data center jobs: Electrical engineer: $100k a few years ago → $500k now Plumber: $65–75k → $300k (& they’re getting poached by nearby sites) The other AI talent war. From @dottkhan Khanference

I'm proud to announce that Proem Acquisition Corp I (Nasdaq: PAAC) is merging with Astro Digital, a premier builder of satellites for commercial, civil, and defense constellations. The combined company is expected to trade on Nasdaq as Astro Digital Holdings, with closing anticipated in Q1 2027. Since 2018, Astro Digital has delivered nearly 40 satellites across 16 mission types for more than 30 customers, including NASA, the Department of Defense, Boeing, and Sony. Its platforms flew the first NVIDIA H100 GPU to orbit on Starcloud-1. The transaction includes a $50M PIPE, with Proem Asset Management and its affiliates committing $25M. Why we chose Astro Digital: Capital efficient: 11 years and nearly 40 satellites without the cash burn that defines most of this sector. Direct leverage to the secular growth of space: as constellations multiply, so does demand for its platforms. Multiple growth vectors: customer follow-ons, new logos, government and sovereign programs, and new categories like orbital data centers. A proven team: Chris Biddy and Michael Wilson built this business from zero. Positive adjusted EBITDA while compounding revenue at 42% annually over two years. Thank you to Chris, Michael, and the entire Astro Digital team for trusting us as a partner on this next chapter. Press release, investor presentation, and webcast: proemacq.com

Some fricking wild stats from our show coming tomorrow with @alexmashrabov: - $1BN ARR in 18 months, faster than Cursor and Cognition. - Higgsfield spend $4M per month on different models - That is over $10K per head. - Alex expects to spend $50-$100K per month on tokens for his 10x people. - One engineer spent $30K in a single week on Astra vibe coding. WTF!! - They have a team of 150 people in content alone. The story that has not been told, coming tomorrow...

I wonder what all these agentic e-commerce orders will do to overall return rates! My guess is return rates will spike, and it will be interesting to see how e-commerce companies manage that increased cost.

Interesting chart from @SVNieuwerburgh’s Brookings paper. US AI capex spending is estimated at over 3.5% of US GDP. For context, capex for railroad buildout between 1870 and 1890 was 2.24% of US GDP. Thank you @MichaelRStrain for pointing this out.

Last month I wrote about how we can build a positive and safe future for everyone: meta.com/thefutureisforevery… Every lab has the responsibility and incentive to move at the pace required to train its models safely, and the ability to take its own actions to ensure that happens. The reality is: - People won't want to use agents that are misaligned with them and that don't do what they ask, so labs have a strong natural incentive to make their models more aligned. There is a lot of debate about slowing progress on capabilities until alignment catches up. My view is that trust and alignment are quickly becoming the most important capabilities that will differentiate agents and models. Any lab that doesn't focus on alignment will fall behind. - Labs face significant liability if their models cause harm, so they have a strong incentive to prevent this as well. Meta delayed shipping Muse for several months to focus on safety and security. We didn't call for everyone else to do this before we would. We just did it as part of our day-to-day work because it was clearly the right thing for people and for us. I'm proud of the security foundations we've built. - Engaging independent evaluators and advisors is industry best practice. MSL already does this today in several areas because it helps produce better work. Other labs can just do this too. In general, it would be helpful for there to be a larger and more diverse ecosystem of evaluators. - Committing the significant majority of compute towards serving people rather than racing towards recursive self-improvement is one of the best ways to ensure we develop this technology safely. Meta has made this commitment and other labs can do this as well. I believe the key to building a positive future for everyone is maintaining the right balance of power. This is within our power to do.

I have a very-out-of-consensus view: The Federal Reserve is unlikely to raise interest rates this week. My baseline: 1. As demonstrated in the July Federal Open Market Committee (FOMC) meeting, the center of gravity of the committee was to hold, not to hike. 2. Several FOMC members seem to share the view of Governor Barr, who said: “At the September FOMC meeting, if trends in the data give me some confidence that inflation is moderating on a path to two percent, then I think we can take a bit more time to assess our policy stance. However, if inflation appears not to be moderating sufficiently, then I think we should act decisively to raise rates.” So the question is whether the August CPI report moved the center of gravity of the FOMC from holding to hiking. It is certainly possible that it did, and I wouldn’t be shocked if the Fed does raise rates this week. But my expectation is that the Fed will hold. Five points: 1. A HIKING CYCLE IS DIFFERENT THAN A ONE-AND-DONE HIKE The Fed rarely does one-and-done hikes. So the right question is not, “Should the Fed raise rates next week?” Instead, the right questions are: “Should the Fed begin a hiking cycle next week?” and/or “Should the Fed do a one-and-done hike, which the Chairman will have to explain as a one-and-done hike?” Viewed this way, the odds of a hike next week should be a lot lower than the 87 percent current priced in by the bond market. The August CPI report might have been enough to move some FOMC members from favoring holding to favoring a one-and-done hike. But the August report would have had to have done a lot more work to move the center of gravity from holding to a *hiking cycle*. (More on the August CPI report below.) Moreover, a one-and-done hike is unlikely due to Chairman Warsh’s approach to communications, which suggests that he would not be eager to explain why the Fed did its first one-and-done hike in decades. I very well might be wrong. It is worth noting that Governor Waller, who is very influential with other committee members, might have signaled support for a one-and-done hike when he said the following on September 3rd: “If there is evidence that progress toward two percent inflation reversed in August, a small adjustment in our stance would help ensure that it resumes.” 2. NOT ROCKING THE MARKET IS A REAL CONCERN Not rocking the market by holding is an argument that many analysts take seriously. If the Fed does hike, then the desire to avoid the market reaction that would follow from holding could be the decisive factor behind the hike. * But the bond market thinks there’s a 75 percent chance of two or more hikes in 2026. So a one-and-done hike would still go against the overwhelming consensus of the market, and a one-and-done hike would still find the Fed rocking the market by going against its pricing. * If Chairman Warsh is looking for an opportunity to signal to the market that the Warsh Fed is different than its recent predecessors, then this is that opportunity. 3. WAS THE AUGUST CPI REPORT THAT TROUBLING? * Based on the August CPI, I'm expecting August monthly core PCE inflation to have accelerated a bit relative to the June and July. But even so, I expect August monthly core PCE inflation to be around 3.1 percent annualized. I am expecting the August six-month change in core PCE to be around 3.2 percent annualized. I am guessing that the majority of the FOMC would view this as evidence of ongoing disinflation (3.1 < 3.2). Or, if not, I am guessing at least that the majority would not view it as evidence of reacceleration that requires a hiking cycle to begin this week. Instead, I suspect that they would prefer to see how the next few months of data unfold. * I suspect that the August CPI print did not change the view of the center of gravity of the committee that, without higher energy prices and tariffs, underlying inflation would be closer to 2.5 percent than to 3 percent. For example, the August CPI report did not show strong evidence of inflation creeping into core services (at least relative to July). In August, core services CPI printed at 3.0 percent over the year. In July, the same measure was 3.0 percent. So if in July you thought that the economy was experiencing some disinflation, and that above-target inflation was largely driven by supply shocks, then the August PCE report we are likely to get would likely not be enough to move you off that view. 4. FINANCIAL CONDITIONS ARE TIGHTENING Over the past couple weeks, financial conditions have tightened, which Chairman Warsh and others on the committee could argue is doing the work that a potential increase in the funds rate was meant to do. 5. TO BE CLEAR, I WOULD RAISE RATES To be clear, my core argument is that not enough has happened since the July FOMC meeting that views should move in either direction. In July, if you thought that the federal funds rate was too low, then you should still think the federal funds rate is too low. And in July, if you thought the federal funds rate was at the appropriate level, then you should still think it is at the appropriate level. My own view is that the federal funds rate is too low and that the Fed should increase the funds rate this week. I would have voted for a rate hike in July (and earlier; I was opposed to the cuts in 2025). The center of gravity of the FOMC has not agreed with that throughout 2026 — including as recently as July. I suspect the data that have come in since the July meeting have not changed their view. I am expecting the Fed to hold this week, though I would not be shocked if I turn out to be wrong.

Law enforcers already have authority to charge companies and their CEOs for creating and releasing dangerous, unvetted, or defective products. We shouldn’t let discussions about new legal regimes distract from the fact that there’s no AI exemption from laws already on the books — a point @FTC emphasized repeatedly during my tenure. 1. There is an extensive set of laws that govern dangerous and defective products. For example, releasing unvetted AI models or agents can violate consumer protection laws. Shipping flawed AI tools without implementing adequate measures to detect and stop rogue or defective AI agents can be an “unfair or deceptive” act or practice under the FTC Act (and analogous state laws). And some state AGs are already exploring holding AI firms and their CEOs criminally liable when their models participate in criminal activity. 2. Existing laws also prohibit “unfair methods of competition.” This covers instances where AI firms appropriate the competitively sensitive information of their customers, including through tracking their use of various tools. It can also cover instances where firms pursue dangerous behavior, aware that doing so may compel rivals to do the same. As the Supreme Court has noted: “A method of competition which casts upon one's competitors the burden of the loss of business unless they will descend to a practice which they are under a powerful moral compulsion not to adopt, even though it is not criminal, was thought to involve the kind of unfairness at which the [unfair methods of competition] statute was aimed." 3. The highly concentrated and interconnected structure of these markets could be creating major risks and conflicts of interest. We had started investigating these partnerships and cross-investments across the stack (and released a preliminarily overview of some findings: ftc.gov/news-events/news/pre…). Both federal and state enforcers should be scrutinizing these opaque relationships and inter-dependencies. We are already seeing how these relationships could undermine accountability. For example, OpenAI could face liability given the Hugging Face incident, but Hugging Face being bought up by Nvidia means that we’re unlikely to see it file a lawsuit over this — given Nvidia’s strong incentive to see OpenAI continue full speed ahead. 4. As AI tools dramatically change the landscape of cybersecurity risks and hacks, all businesses should be doubling down on having core security protections in place. Firms that fail to invest in adequate data security measures or fix known vulnerabilities can also be breaking the law. A recent analysis showed that around 1/3 of Fortune 100 companies do not even have a way to notify them about security issues. During my @FTC tenure, we sued firms for poor data security practices and held CEOs liable when they were personally responsible. this.weekinsecurity.com/doze… ftc.gov/news-events/news/pre… 5. As policymakers consider new legal regimes, we should be looking to lessons from prior efforts to govern major sectors, such as banking and other networks, platforms, and utilities. Tools like structural separations, nondiscrimination, and supervision could be key, and there’s a rich history of what works and what doesn’t. But we can and must pursue any new efforts alongside enforcing existing laws.