- Welcome to *. Cautious Optimism, a newsletter on tech, business, and power. Modestly upbeat
Friday. It’s Juneteenth here in the United States, a holiday that commemorates the day Union forces arrived in Galveston, Texas, ensuring the freedom of enslaved people in the state. June 19 is celebrated to mark emancipation in the United States more broadly and became a federal holiday in 2021 during the Biden administration. Happy Juneteenth!
- Never forget that the Confederates were not merely traitorous racists. They were traitorous racist losersto boot.
Today we’re looking at Kalshi’s exploding revenue, nuclear power, and how a Chinese company just checked **Google **in the AI game. To work! — Alex
📈 Trending Up #
The ceasefire? …Microsoft, the European company…SpaceX recycling debt…catching strays…self-driving IPOs…Jio’s IPO! …early-stage M&A? …public corruption? …
What’s the latest with Anthropic’s Fable model? The company is working with the White House, Politico reports, to create a “framework that would assess the severity of security flaws in new AI models and guide potential government intervention.” In the end, the technology-right wound up with a President willing to force the takedown of AI labs while creating novel regulations to control the release of future artificial intelligence technology. Well done!
- After concern that the Trump administration didn’t understand the difference between deterministic and probabilistic systems, the same Politico story reports that current talks between the AI lab and the government “reflect an understanding that no AI model can be completely immune to hacking.” Praise be.
Kalshi’s revenue: The Information reports that Kalshi has reached a $2 billion run rate and is holding “informal IPO talks.” Since prediction market operating data is pretty darn public thanks to Dune, I took a peek. Here’s what @datadashboards came up with:
May saw calculated Kalshi fees reach $159 million ($1.91 billion run rate); June has seen $128 million thus far, on pace to top May if the pattern holds. And it may. The NBA Finals were a big deal at the start of June, and now that the contest has concluded, the World Cup is in full swing. (More than half of Kalshi’s transactions at present are sports-related, mind.)
Prediction markets enjoyed a period in the sun during the last American election; current sporting events provide more fodder for wagering. The question before Kalshi, Polymarket, and their rivals is how well they can weather periods of low sport. Probably pretty well? People really do love to gamble.
**Nuclear power: **As we keep tabs on the global compute buildout, a key topic is power. What’s great at making lots of power? Nuclear energy. Hence, our interest in Standard Nuclear’s upcoming IPO. The nuclear fuel maker is deeply unprofitable, but the timing of its public offering suggests an expectation strong investor demand for shares in companies that may prove critical in the near future.
- Standard Nuclear makes tristructural-isotropic (TRISO) nuclear fuel, the very substance that Valar Atomics wants to use as fuel in its SMRs. Synergy! And Valar is in the news this week for taking itsWARD 250reactor to criticality for the first time (the startup’s second criticality event). Hell yeah, nuclear power rules, and we need more of it.
[📉](https://finance.yahoo.com/news/servicenow-pledges-1-5bn-investment-110000403.html) Trending Down
[📉](https://finance.yahoo.com/news/servicenow-pledges-1-5bn-investment-110000403.html)
Sherwood…Amazon’s ability to have a spine…Meta’s available compute, incredibly…free speech in India…using AI in school…
**Google’s AI prowess: **In late 2025, Google released Gemini 3. At the time, the model was considered such a strong entrant in the AI market that it put OpenAI on its heels.
How times have changed. Today, Google’s Gemini 3.1 Pro Preview is ranked as the 7th most intelligent model, while its more recent Gemini 3.5 Flash ranks but fifth. Incredibly, both of Google’s latest AI models are now behind the recent release of GLM-5.2, created by Chinese AI lab** Z.ai**.
Not only is GLM-5.2 cheaper than Gemini 3.5 Flash (and 3.1 Pro Preview), but Artificial Analysis ranks it as the most intelligent model in the world after Claude Fable 5 (#1), Claude Opus 4.8 (#2), and GPT-5.5 (#3). To see an AI model of GLM-5.2’s power released under an “MIT open-source license” with “no regional limits” is insane.
- Perhaps the main risk to American AI labs is not that China catches up to the frontier, per se. But merely that its AI model makers skate so close to the performance edge at a lower price point that they collapse the ability for Anthropic and OpenAI to charge top-dollar for that last squeeze of intelligence. Maybe. What is entirely clear, however, is that Google is losing its grip on the AI bronze medal.
**American-Italian relations: **Here’s a crazy story. Once upon a time, Italy’s Prime Minister Giorgia Meloni and President Trump were buds. Meloni, once a youthful ~neofascist, is today a conservative populist. You can see why she and Trump would get along. From the start of Trump’s second term through early 2025, Meloni and POTUS were in sync. Then, trade tensions and differences in perspective on Ukraine (Meloni supports Ukraine, while the American executive keeps trying to convince the nation to cede land to Russia) shot fractures through the relationship.
This year, Meloni and Trump have been at odds over the USA-Iran war, with Italy not allowing the United States to use its land as a base for combat operations. POTUS didn’t like that. And then, this:
The spat between Donald Trump and Giorgia Meloni escalated dramatically Friday, after the U.S. president
[said]the Italian prime minister “begged” him for a photo at the G7 summit earlier this week.During a phone call with Italian media on Thursday night, Trump said Meloni was “probably happy I spoke to her. I didn’t have to,” adding that “she wanted a picture with me so badly. I wouldn’t have taken it, but I felt sorry for her.” The remarks appeared to contradict the prime minister’s recent insistence that her relationship with the president was as good as ever.
Meloni called Trump’s statement “completely made-up” in a combative video on social media, saying “Italy and I never beg.” She added that she was “stunned” by the president’s remarks, and that “even though this is not happening for the first time … it’s a pity that [Trump] does not show the same determination against the enemies of the West.”
I mean, look, even the European right has found its spine, viz Trump and his bloviating. It’s notable how the Iran conflict not only further harmed POTUS’s domestic approval ratings, but has seemingly damaged his (and, by extension, my nation’s) international standing.
Great work, everyone.
How would a national AI policy even work? #
The Vice President is currently on a book tour, giving him ample microphone minutes to chat about a host of topics. Including AI. Over on The Diary of a CEO, Vance riffed about AI, faith, wealth inequality, and more. What follows are excerpts of his remarks, pulled from a YouTube transcription and *lightly *cleaned up for your reading pleasure. (Check my work; watch the original clip; I did not even try to add every single required comma.)
** JD Vance on AI-induced job loss:** [After discussing deindustrialization of the United States and it being described as inevitable by some:] I think sometimes we tell ourselves a story that technology always leads inevitably to job loss to make up for the fact that what often leads to job loss among populations is either outsourcing or immigration. You ship the job to another country, or you have somebody else take the job of somebody who currently has it.
So what do I think is actually going on with AI? I think you know if you go back to the Industrial Revolution, the last significant major disruption in the labor market, you actually had way more people working at the end of the Industrial Revolution than you did beforehand. Again, some of the jobs were different. There was some job disruption. But when I look at AI, I don’t see mass unemployment as the most likely consequence. I think people will become more productive. I think some people’s jobs will change. Some people will lose their jobs. But I just don’t buy this idea and I haven’t seen any evidence in the data that it’s going to lead to mass unemployment.
The question of AI-driven job loss remains unsettled. Recent college graduates complaining about a weak labor market and a decline in corporate interest in hiring junior staff in certain roles stand in contrast to relatively low overall unemployment.
JD Vance on the risks of wealth concentration, and AI-predicated political unrest:
Let me tell you what does what does worry me. Again, historical analogies are always fraught. You go back to the Industrial Revolution: Was mass joblessness the main consequence of the shift from an agricultural to an industrial economy? No. But what did happen? Rich people got way richer. And that led to in Europe fascism and communism. In fact, your country [the host is British] and my country [were] pretty much the only two countries that successfully avoided either a fascist or a communist revolution in response to the industrial revolution.
That’s by the way one of the real interesting things about you know Christianity is the the seminal text about how capital and labor could work together about how you could have social harmony. compared to Marx. who was sort of saying there’s an inevitable social division, was Pope Leo I 13th who wrote in his famous encyclical that the way to preserve harmony between the social classes was to ensure that the workers could bargain. This is where sort of the idea of collective bargaining had a Christian underpinning and to make sure that the the capitalists weren’t able to take advantage of the workers.
They had to sort of respect them. And that that that model of social harmony I think is something we’re either going to follow that Christian concept of social harmony in the age of AI or we’re going to wake up and we’re going to realize that rich people have gotten way richer. The average American, the average Brit, the average Western society member has stagnated and people really hate relative poverty. You can give people iPhones and you can give the people the creature comforts of a 21st century economy, but if you make rich people way richer, you are going to have significant problems. And so I think that is like one of the consequences that I see from AI.
VPOTUS here argues that collective bargaining is inherently Christian and that allowing capitalists to control too much power is a recipe for either left-wing or right-wing revolution. You could read this comment as a condemnation of unvarnished capitalism, though I think that Vance is more arguing that with proper guardrails, AI won’t necessarily push the United States in either direction.
JD Vance on what about AI makes him worried, continued:
The one other [concern] just because you asked about it, the other thing I really worry about with AI is surveillance. AI is, you know, a friend of mine once said that AI is fundamentally a communist technology in that it allows governments and corporations to surveil people in very profound and different ways. And that scares me a lot. Like I don’t want a social credit system that’s powered by AI. I don’t want you to not be able to buy a beer because some tech CEO has given you a score based on an artificial intelligence algorithm that nobody actually understands. That scares me, too. But I don’t think we’re going to have mass unemployment. We might have mass inequality. That’s its own problem.
VPOTUS appears to quote Peter Thiel here, who said back in 2018:
Crypto is decentralizing, AI is centralizing. Or, if you want to frame it more ideologically, crypto is libertarian and AI is communist.
It’s a bit rich to warn about wealth inequality in one breath and then cite one of the world’s wealthiest individuals in the next. But, hey, let’s allow people friends. The ~good news is that VPOTUS sees a path forward that allows for societal oversight of AI without draconian rules. He gives an example:
There is a very deeply Christian concept that you have to give everybody in the country a seat at the table. So for example like okay there’s the economic piece of it. What about the cultural piece of it? How will AI transform the culture that we consume that we distribute that we make? You know, back in the 50s and 60s, it was broadly accepted that now it wasn’t a censorship regime. There was nothing legal going on here, but it was broadly accepted that Hollywood would consult with the religious leaders at the time in order to ensure that the content they were making was actually consistent with the sensibilities of their membership and consistent with some basic Christian ideas. Again, that wasn’t forced, but there was this mechanism that gave everybody a seat at the table. And I think that’s one of the bad things about the there are many bad things about the decline of institutional Christianity in this country. But we do not have a mechanism that gives powerful people that forces them to actually work with everybody else.
Religion was one of the ways that happened in the West. I think probably the most profound and effective way that happened in the West. We just don’t have it anymore and I really worry about that.
*Vance, I believe, is referencing the Protestant Film Commission, for what it’s worth. I don’t know how much I like the idea of granting religious bodies power over the central government when it comes to making the sort of choices in question; in fact, handing religious leaders the con is probably the only thing worse than giving it to the government. *
JD Vance on partial nationalization of AI labs:
[The President] likes the idea as sort of a sovereign wealth fund idea of the United States taking some stake in these AI companies. He said so publicly. I’m not breaking news. But you know again the president he is a very unconventional person. You would say a Republican is not supposed to think like that. The president doesn’t care. The president just thinks the thoughts that he has. He develops them whether they’re, you know, he tries to determine is this a good idea or a bad idea. I would call him sort of a radical pragmatist, though I think most Europeans think that he’s this hyper ideological person. He’s extremely pragmatic about this stuff. But but but one one one very important thought. The idea that we’re going to allow these companies, let’s say 10 20 years down the road to accumulate trillions and trillions and trillions of dollars of wealth and then we’re going to be be able to successfully redistribute it to workers. I’m very skeptical of that.
This is softly spoken, but it appears to be either VPOTUS creating room for Republicans to support partial government takeover of AI labs, or at least him attempting to gin up air cover for Trump to do things that he disagrees with.
In sum, VPOTUS is not worried about AI-driven job loss but wants to balance the power of capital and labor while incorporating community input into AI choices *and *possibly giving the aggregated public a stake in (currently) private equity.
Without the Thiel reference, you’d almost think the Vance was a bit left-of-center. He’s not, but here we can see how Christian teaching at times doesn’t mesh with present-day right-wing orthodoxy. And yes, we’re being ironic twice.