Good morning. Andrew here. We’ve got news this morning on a new open-weight A.I. model that could challenge Anthropic and OpenAI — and it’s from an American start-up. The implications of a successful U.S. open-weight model are significant. If companies move away from more costly closed models, the economics of the business models of Anthropic and OpenAI fundamentally weaken. That would batter their valuations, and could also hit those of other big tech giants, many of which have invested in the labs. More below. (Was this newsletter forwarded to you? Sign up here.)
A new open-weight A.I. contenderAnthropic is expected to go public next month in perhaps the biggest I.P.O. in history. Yet the artificial intelligence giant is facing a growing set of headwinds. There’s the question of how safe its models are (more on that below). And then there’s increasing competition, not only from OpenAI, but also from cheaper open-weight models from China — and now from Reflection AI, an American start-up backed by Nvidia, Sri Muppidi reports. Meet Beam, Reflection’s long-awaited model. Beam, introduced yesterday, is focused on coding and A.I. agents, applications popular with business users. Reflection claimed that Beam’s reasoning performance is on par with GLM-5.2, a powerful open-weight model by the Chinese start-up Z.ai, and outperforms many Western open-weight ones.
It’s the first volley in the A.I. race by Reflection, which was founded in 2024 by two former Google DeepMind researchers, Misha Laskin and Ioannis Antonoglou. The company has raised billions of dollars from investors including Nvidia, Sequoia, Disruptive Technology Advisers and 1789 Capital, the investment firm at which Donald Trump Jr. is a partner. Reflection was most recently valued at $25 billion. Open-weight models have become a big topic of debate in A.I. These tools, which users can modify, have risen in popularity as companies seek cheaper alternatives to closed-source models from places like Anthropic, Google and OpenAI. Developers have increasingly adopted powerful open-weight models from Chinese labs, worrying leaders in Silicon Valley and in Washington. That has stoked interest in similar tools developed by Western companies like Meta; Thinking Machines Lab, which a former OpenAI executive founded; and Mistral, France’s A.I. champion. Reflection adds to the challenges confronting Anthropic, as the company marches toward its I.P.O.:
The Supreme Court considers whether oil companies can be sued over climate change. Justices yesterday heard arguments over whether localities and states can sue energy firms to pay for damage they say extreme weather caused. (They appeared to be skeptical of industry attempts to shut the lawsuits down.) If the case proceeds, it could open the door to dozens of similar claims. President Trump pushes to lower diesel prices before midterm elections. Trump signed an executive order allowing drivers to buy tax-exempt “red-dyed diesel,” which is mostly used in agriculture and construction. The average price of diesel remained around $6.32 a gallon, as strikes on refineries in the Middle East and Russia weigh on worldwide supplies, even as Brent crude, the international benchmark for oil, dipped below $98 a barrel today. The Fed is reportedly reviewing Wall Street banks’ private credit loans. Officials from the New York Fed have examined JPMorgan Chase, Wells Fargo, Barclays and Morgan Stanley since the spring, asking questions about their exposure, risk management and collateral, Semafor reports, citing unnamed sources. One motivation for the review was JPMorgan’s decision in March to mark down some loans to private credit lenders, according to Semafor. Elon Musk is a trillionaire again. A surge in SpaceX and Tesla shares yesterday pushed Musk’s net worth up $65 billion, to $1.04 trillion. His jump in paper wealth comes as artificial intelligence propels the fortunes of some moguls: The roughly 100 technology-based fortunes on Bloomberg’s billionaires list gained a combined $845 billion in paper wealth for the first nine months of the year, while non-tech ones lost a combined $62 billion over that period. How long can Wall Street’s profit bonanza last?Wall Street had a stellar first half of 2026 profit-wise, helped by a bumper period for deal making, buoying the fortunes of its hometown. The question is whether a potential slowdown in mergers and I.P.O.s could throw cold water on the good times. Exceeding expectations: Profits at Wall Street banks reached $45.9 billion for the first half of 2026, according to a report released this morning by Tom DiNapoli, New York State’s comptroller. That’s up about 51 percent over the same period last year. More from the report:
What’s behind the bonanza: In a statement, DiNapoli cited the boom in artificial intelligence spending (now fueled by huge borrowing that banks are arranging) as well as increased trading volumes because of high market volatility. M.&A. was also a big contributor. The value of deals announced worldwide in the first half of the year was $2.77 trillion, according to LSEG, the highest amount during the first half of a year since 2002. But there’s reason to worry about the rest of the year. M.&A. volume declined sharply in the third quarter, which saw the lowest level of transactions worth more than $10 billion since the fourth quarter of 2024. Deal makers cited rising borrowing costs and growing inflation as causes for concern, even as many still expected a strong finish to 2026. Another big question is I.P.O.s, as companies like the health device maker Oura delayed planned stock market debuts. A growing number of I.P.O.s have been postponed as bond yields remain high, making investors nervous about buying into risky equity offerings. (I.P.O.s are more sensitive to market conditions than mergers.)
VIDEO OF THE DAY A stark A.I. warning for New York lawmakers
Jacob Coxon, the researcher whose warning about the existential risks of artificial intelligence set off a global firestorm last month, testified yesterday before the New York City Council. “On the current path,” Coxon said, “I think it is more likely than not that humanity loses control to these A.I.s and it could end in human extinction.” His remarks echoed his social media post announcing his resignation from Anthropic over concerns about the pace of its artificial intelligence development. That post now has nearly 175 million views. Coxon’s warning “really spurred many levels of government into action,” Assemblyman Alex Bores told DealBook. Bores, who ran for New York’s 12th Congressional District this year with a campaign focused heavily on A.I. regulation, also testified at the hearing yesterday. The City Council is weighing roughly 10 bills targeting A.I. They include regulation that would require companies to receive an independent safety evaluation of their models before releasing them and provide protection for whistle-blowers at tech companies. Lawmakers in the state are also planning to introduce a “ban on superintelligence or recursive self-improvement,” Bores said. That’s starkly different from President Trump’s preferred approach for letting A.I. companies self-regulate. The A.I.-savviest banksFor most industries, racing to adopt artificial intelligence is considered important, if not increasingly existential. That’s especially true in banking. The research firm Evident said deployment of the technology in that industry had hit its fastest pace since 2023. Evident also laid out its latest ranking of how banks are adopting A.I. — and told Michael de la Merced what lenders are increasingly watching out for when it comes to A.I. tools. The methodology: Using public data points, Evident assessed 50 major banks in North America, Europe and parts of the Asia-Pacific region on four qualities of A.I. adoption: talent, innovation, leadership and transparency. Here’s the top 10 and where they ranked last year:
Why it matters: The leaders in the rankings were all early A.I. adopters that invested heavily in the technology, setting them up to be more efficient and create better tools, Alexandra Mousavizadeh, a founder and a C.E.O. of Evident, told DealBook. (Morgan Stanley, for instance, has saved more than one million developer hours with A.I., which it’s reinvesting in technology, according to Evident.) By contrast, Mousavizadeh said, laggards could start to see their A.I. adoption shortfalls hitting their businesses in the next year or two. They could also face tougher competition from more tech-savvy rivals, including from younger lenders like Revolut.
Banks are keeping an eye on two things related to A.I.:
We hope you’ve enjoyed this newsletter, which is made possible through subscriber support. Subscribe to The New York Times.
Deals
|
