TheChek News Flow

The Philadelphia Police Department said the agents had also sent in a false homicide tip. The incidents led the White House to call for better disclosure of rogue A.I. behavior.

Artificial intelligence agents created by Anthropic, acting on their own, tried to gain access to a range of government websites, filled out visa applications on the State Department’s website and sent a fake homicide tip to Philadelphia’s Police Department.

Anthropic detailed the activity of its A.I. agents in a blog post on Friday, without naming the targeted websites. But two sources with knowledge of the incidents said Anthropic’s A.I. agents had submitted 20 visa applications through a form available on the State Department’s website. All the applications were incomplete and were not processed, they said.

White House officials said they were briefed on the agents’ actions on Friday, and demanded that A.I. companies immediately report rogue A.I. activity. It was the most aggressive statement so far on A.I. regulation by the Trump administration, which has sidestepped calls for tighter controls on the increasingly powerful technology.

A new White House A.I. task force, created this week, said Anthropic told it about the agents’ activities earlier on Friday. Called the Super Intelligence Force, the group includes Jay Clayton, the White House A.I. czar; Andrew Ferguson, the chair of the Federal Trade Commission; Scott Kupor, the director of the Office of Personnel Management; and Emil Michael, the chief technology officer of the Defense Department.

“We informed the company that we expect immediate and full transparency to the entities involved and the public,” they said in a statement. “We also expect that the company will immediately provide remediation services to the affected entities and any harmed Americans.”

They added that they expected Anthropic and other companies to “immediately report incidents,” cooperate with federal and state law enforcement and ensure that the incidents were not repeated.

Anthropic said in its blog post that it discovered what its A.I. agents had done after it started a review of its A.I. activity in July. At the time, OpenAI had disclosed that its technology attacked another A.I. start-up, Hugging Face. Anthropic and other labs also discovered that their technology had escaped from testing environments and conducted hacks.

Anthropic said an A.I. model it had been testing had taken several unauthorized actions, including using a flaw in a university website to download data and sending a government agency a form that it had been told not to submit.

A spokesman for Anthropic declined to comment beyond the company’s blog post.

The Philadelphia Police Department said earlier on Friday that Anthropic had notified it that the company’s technology submitted a false homicide tip to the agency’s website. The tip was dated July 18 and claimed that the submitter had information about an unsolved case, the department said.

The tip was flagged as spam and never investigated, the department said, but Anthropic notified the police this week about the incident and said the company would publicly disclose it on Friday.

“An unreleased, non-frontier research model was meant to fill out a practice copy of a government form,” Anthropic said in its blog post. “When the copy failed to load or the model closed it by mistake, the model instead navigated to the website where the real form is normally hosted and submitted the form there.”

Public concerns about A.I.’s going rogue have risen as more artificial intelligence labs have reported losing control of their technology. Google, Meta, OpenAI and Anthropic have been under scrutiny for the dangers that cutting-edge A.I. models might pose. They also face questions about whether to slow down development of their systems.

(The New York Times has sued OpenAI and Microsoft, claiming copyright infringement of news content related to A.I. systems. The two companies have denied those claims.)

Safety experts have warned that rogue A.I. could take down critical infrastructure like electric grids or banking systems. After A.I. models from several leading labs broke out of testing environments this summer, the companies said they planned to introduced enhanced safety measures to control them.

The expanded term would encompass event contracts on sporting, political, cultural and weather-related events

The Commodity Futures Trading Commission is proposing to provide what it describes as additional clarity on the term “swap” that would include event contracts used in prediction market trading.

The expanded term would encompass event contracts on sporting, political, cultural and weather-related events, the CFTC said Friday.

The proposal notes that event contracts are financial instruments commonly known to trade as swaps.

The prediction markets industry, which is regulated by the CFTC, has been enmeshed in a debate about whether trading on events constitutes gambling, which would put it under the legal purview of states. The CFTC has said that the emerging industry is a form of financial trading, not betting.

“These products are commodity derivatives squarely within the CFTC’s regulatory remit under the Commodity Exchange Act and are within the agency’s exclusive jurisdiction,” CFTC Chair Michael Selig said.

The agency is seeking public comment on the proposal. Comments must be submitted within 30 days.

The Desperate Hunt for AI Computing Power Is Upending Silicon Valley

Bitter rivals are forming alliances and executives are having to personally intervene over the scramble for resources to fuel the AI boom

Google really wanted pioneering AI researcher Noam Shazeer back.

The longtime employee left to start his own company but in 2024, Google paid $2.7 billion for him to return to work on cutting-edge projects, including its most powerful AI technology, Gemini.

Even that, though, couldn’t ensure that Shazeer got the computing resources he wanted.

Less than two years later, he sent a note to one of his teams saying that Google had cut what was allocated to them in favor of another project. Shortly thereafter, Shazeer left for rival OpenAI.

Compute, the industry shorthand for computing power, is the lifeblood of the AI boom, and it’s in short supply. Now fights over how to get more of it—and what projects to give priority to—are sending a powerful industry scrambling. Bitter rivals are forming new alliances, such as Elon Musk and Dario Amodei, whose companies worked together to secure computing power. Venture-capital firms are touting access to AI chips to woo startups. And power struggles are more frequent as companies compete in a worldwide treasure hunt for the tech industry’s most coveted resource.

“There’s a big gold rush happening at the moment,” said Evan Conrad, chief executive officer of San Francisco Compute. “It’s hard to fully comprehend the scale.”

Conrad’s company vets data centers and then rents their computing power to AI companies that range from the largest players like Nvidia to small startups like Standard Intelligence. Its chief technology officer, Eric Park, travels the globe in search of computing resources hiding where few else think to look.

In the spring, it was far-flung corners of rural America, where one promising new data center turned out to be AI servers humming inside what appeared to be a converted chicken coop. Then Manila in monsoon season, where Park’s car stalled in floodwater and he and his colleagues waded to safety, bags held over their heads. Eventually, a passing truck shuttled him to the airport, where he bought a new pair of shoes and boarded a flight to Taipei—to meet with a company building servers for the AI boom.

Even a little cluster of available AI chips can be helpful to startups trying to train or fine-tune smaller AI models.

SF Compute’s business has grown as companies are increasingly beholden to the yearslong timelines of data-center build-outs. There’s a race to build more data centers to get more capacity online fast, but construction is stymied by regulation and political backlash.

Established suppliers of computing power are requiring customers to sign multiyear contracts with heftier down payments. Startups are increasingly inking deals with unreliable providers out of desperation, like ones lacking proper security or weaker infrastructure.

While splashy announcements have promised enormous amounts of new computing power coming soon, far less is actually available today, said Nick Frosst, co-founder of Canadian AI startup Cohere.

Advertisement

“Lots of things have been announced and then canceled,” said Frosst, whose startup builds large language models and was valued at $6.8 billion last year. “Or things that they said they’re going to build this big, but then they actually build significantly less.”

Arcee, an open-source AI startup, recently secured commitments for both capital and computing capacity from an infrastructure partner, said CEO Mark McQuade, but the deal collapsed when the firm’s data center was delayed.

“So much is up in the air with these things, especially when it comes to compute,” said McQuade. “You take it with a grain of salt until it’s real.”

With so much demand, prices are going up. The hourly cost of renting computing power from Nvidia H100 chips on a one-year contract has risen 60% in the past year, according to SemiAnalysis, a data and consulting firm. Alphabet GOOGL 0.97%increase; up pointing triangle expects to make up to $200 billion this year in capital expenditures, much of it going toward AI infrastructure.

Even the busiest AI executives are personally involving themselves in negotiations over computing power.

In late March, Anthropic co-founder Tom Brown visited Elon Musk at rival AI company xAI’s offices in Hawthorne, Calif., to broker a deal to rent computing power from Musk’s company, according to people familiar with the matter. The explosive, unexpected success of Anthropic’s Claude Code had created a computing crunch, contributing to frequent outages.

In May, Anthropic stunned the industry when it announced an agreement with Musk’s SpaceX to lease more than 300 megawatts of computing power—part of a deal later revealed to be worth up to $45 billion in the coming years. Musk, who had previously been publicly critical of Anthropic, posted on X that he was “impressed” with members of Anthropic’s senior team, adding, “No one set off my evil detector.”

Anthropic CEO Dario Amodei also spoke on the phone with Meta chief AI officer Alexandr Wang earlier this year, attempting to source more computing capacity, according to a person familiar with the matter. Meta debated whether to give chips to Anthropic and ultimately decided not to for now, people familiar with the matter said.

This is all in addition to the hundreds of billions of dollars in computing deals Anthropic has announced over the past year.

Inside Google’s Mountain View offices, researchers are fiercely divided over which projects deserve the company’s precious computing resources. Co-founder Sergey Brin sometimes upends the formal compute allocation process, reshuffling power around to projects that he thinks are important, according to people familiar with the matter. Demis Hassabis, who co-founded and until recently ran Google’s AI lab DeepMind and took a chairman role in August, has expressed frustration about a lack of compute allocated toward long-term research, according to people familiar with the matter.

To hedge against the unpredictable nature of securing computing power, some companies are signing contracts for more capacity than they need. In other cases, startups have predicted surging demand for their AI products that never materialized, forcing them to resell leftover capacity.

This confluence of factors coupled with rising wait times and skyrocketing costs of compute are crippling a generation of newcomers trying to compete with Google, Anthropic and OpenAI.

Earlier in the AI boom, companies large and small could rent clusters of AI chips from Amazon Web Services, for example, without having to make a multiyear commitment or pay an astronomical upfront fee.

As frontier AI labs soaked up the available compute, cloud providers moved to require multiyear contracts with upfront costs often as high as 30%. Sometimes they require startups to find guarantors willing to cover their payments if they are unable to.

“It’s like if you had an apartment, and you could rent month-to-month, and then suddenly your landlord comes in and says, ‘Hey, I need you to sign a five-year lease, and by the way, I need you to pay a year upfront,’” said SF Compute’s Conrad.

The challenge isn’t only securing computing power but accurately predicting how much firepower startups will need as they scale.

One of Silicon Valley’s newest and fastest-growing startups, the viral AI assistant Instinct, is dealing with steep competition from Meta and OpenAI, which both recently debuted their own personal AI agents and have access to substantially more capital and computing power.

Instinct last month raised $1 billion at a $10 billion valuation in what was its third capital raise this year, but even with oodles of cash, it may struggle to keep pace with the likes of Meta, which owns and operates its own data centers.

Instinct CEO Noah Shinn said on a recent podcast episode of “Invest Like the Best” that he has wrestled with whether to buy compute now to accommodate a future where it has 100 million users, a risky move that could tank the company if it doesn’t grow as expected. A representative for Instinct declined to comment.

“If you overbuy, you could bankrupt the company by having too many lease commitments, and if you underbuy, you could knock down your growth trajectory and have a bunch of unhappy users,” said Zach Bratun-Glennon, a general partner at Gradient Ventures, which invests in AI startups.

Some venture-capital firms are now reserving clusters themselves and renting the capacity back to their startups, or offering to guarantee contracts with compute providers. Last year, one of Y Combinator’s startups needed 120 graphics processing units, or GPUs, for a two-month training run but couldn’t find a suitable contract, according to a person familiar with the accelerator. So Y Combinator introduced a cluster dedicated to its startups, which it launched in July.

At Radical Ventures, an AI-focused VC firm that recently backed a new AI lab co-founded by Alphabet’s former chief scientist Jeff Dean called Discovery Loop, David Katz leads a team that tracks available GPUs, monitors pricing and contract terms, and connects portfolio companies with compute providers.

Over the past five years, Katz and his small team have negotiated billions in compute deals, he said. Katz often finds himself texting directly with the leaders of neoclouds—companies that purchase GPUs and other chips then lease them to customers—to inquire about availability. He flags potential investments to compute partners before Radical has decided to invest to make sure capacity will be there if they need it.

“It used to be, ‘Where am I getting my capital?’ Now it’s like, ‘Who’s going to help me?’” said Katz. “There’s a lot of capital floating around for these companies, but there’s not a lot of deep intelligence on how to build these things.”

Anthropic AI Model Went Rogue, Submitted Fake Unsolved Murder Tip

The AI giant disclosed new incidents involving websites run by U.S. government agencies at the federal, state and local levels

Rogue AI models have hacked companies and tried to break into government websites. Now one has submitted a fake tip for an unsolved crime.

Police in Philadelphia say Anthropic alerted them this week that one of its artificial-intelligence models submitted a false homicide tip through PhillyUnsolvedMurders.com.

In a blog post, Anthropic said the fake tip was among a group of rogue actions the company uncovered recently. The rogue activity also involved websites run by U.S. government agencies at the federal, state and local levels. The company said it has briefed the White House on these cases and notified each agency involved, and plans to report new instances of unintended AI behavior as it uncovers them.

“While these cases had minimal impact, we do not want to diminish the findings, because the same behaviors could do far more harm as models become more powerful,” Anthropic said in the blog post.

The false-report incident happened in July. Anthropic didn’t realize it until September.

“The City of Philadelphia takes this incident very seriously,” the police department said on Friday. “The two-month delay in detecting and reporting the incident to the city is unacceptable.”

Anthropic’s model was conducting a test involving interactions with randomly selected websites when it submitted the tip pretending to be someone who might have information about an unsolved case, police said.

The submission, dated July 18, went to spam and was never sent for further investigation. That doesn’t take away from the seriousness of what happened, police said.

“Unsolved cases involve real victims, grieving families and investigators working to secure answers,” the department said. “Technology companies must take all appropriate steps necessary to prevent their systems from submitting false information to law enforcement.”

Anthropic told the Philadelphia police that the company discovered the incident on Sept. 28, and terminated the automated testing process responsible for the submission, the department said. Anthropic also told the police it added an additional validation mechanism for future testing, according to the agency.

Anthropic notified the Philadelphia police on Oct. 7, nine days after discovering the incident. Police said the company met with the department the next day.

Anthropic said it identified most of the cases disclosed Friday in a review of transcripts that the company began in July.

Multiple times, the company said, an unreleased model was directed to fill out a practice copy of a government form but navigated to the website where the real form was hosted and submitted the form.

The false murder tip originated when the model encountered a tip form run by a police department on a page referencing an unsolved homicide. The company said Claude was instructed never to log in, create accounts, enter personal data, make purchases or submit anything destructive, but said the instructions didn’t rule out form submissions.

Claude entered: “I may have information regarding this case. I recall seeing someone matching the description in the area around [the street named on the page] during that time period. Please contact me if this information is relevant.” Anthropic said Claude left the name and contact fields empty.

Anthropic said it has built tools to automatically detect and block these kinds of behaviors.

Anthropic has previously alerted police departments to more substantive information about potential risks. In late September, the sheriff’s office in Florida’s Lee County arrested a woman after a tip from Anthropic. The woman had made statements that she would “shoot up” the county’s sheriff’s office, according to the arrest report.

AI companies haven’t always alerted law enforcement to potential risks. In February, Canadian law enforcement named Jesse Van Rootselaar as the suspect in the Tumbler Ridge mass shooting, months after OpenAI employees had considered alerting law enforcement about Van Rootselaar’s interactions with the company’s chatbot. OpenAI ultimately decided not to contact authorities, and the company now faces a lawsuit from British Columbia as well as dozens of victims.

PhillyUnsolvedMurders.com, where the fake tip was sent in, is a Philadelphia police website that lists unsolved homicides, takes tips on cases and offers a $20,000 reward for information that leads to an arrest.

“Members of the public should continue to submit legitimate information about unsolved homicides through PhillyUnsolvedMurders.com,” the police said.

While the chances of a lab accident triggering an epidemic are virtually zero, the disease remains an active threat in many parts of the world

The death of Russian plague researcher Daria Shipilova has ignited a firestorm of speculation over the circumstances of her death and a chorus of concern over the potential for an epidemic. While details of the incident remain scant and contradictory, history offers us some important context.

Fear is an understandable initial response to a pathogen so odious it has acquired the moniker “plague”. The Yersinia pestis bacterium was responsible for two of the deadliest pandemics in human history; the plague of Justinian (541-549 AD) and the Black Death. By the end of its terrible reign in the mid-1300s, the Black Death had killed somewhere between 30 and 60 per cent of Europe’s population.

Infections can take a variety of forms, depending on how the disease affects the body. Most common is bubonic plague, defined by swollen, blackened lymph nodes that gave the middle-age pestilence its name. Septicaemic plague results from bacteria entering the bloodstream, while inhalation of bacteria leads to pneumonic plague. While severity and prognosis vary markedly, the same agent is responsible for all forms.

Today that once relentless killer is now relatively easily treated with antibiotics, its deadly legacy often considered a relic of antiquity. But Y pestis never went away — it remains endemic in parts of the world. Between 2019 and 2025, the WHO recorded 3,847 suspected cases and 423 deaths worldwide. While most stemmed from the Democratic Republic of Congo and Madagascar, there were also cases in the US and China. Russia itself last reported wild cases in 2018. As plague remains an active threat in many regions, Y pestis is still actively researched worldwide. Plague has been endemic in Siberia for millennia — the Irkutsk Institute where Shipilova worked was established there in 1934.

As with all research, there are risks. Pathogen research is conducted at an array of universities and hospitals worldwide, and laboratory-acquired infections (LAIs) and accidental pathogen escape from laboratory settings (APELs) are periodic occurrences. Historical incidents of lab infections have involved everything from poliovirus to Ebola. A review in the Lancet Microbe journal identified 309 LAIs and 16 APELs worldwide between 2000 and 2021. Of the eight LAI fatalities during the period, plague exposure accounted for two. Crucially, however, even with highly pathogenic agents, neither LAIs nor APELs typically result in onward chains of transmission due to multiple levels of containment and risk mitigation.

Even so, Russia’s long history of being less than transparent about medical hazards renders it harder than it should be to clarify matters. While Moscow is obliged to report such incidents within 24 hours, the incident came to the attention of WHO only after conflicting stories were being shared on local and social media. One version insists Shipilova was exposed after breaking a test tube, concealing the accident and fatally delaying vital life-saving measures. Another telling asserts she was exposed during a field trip. Russian health authorities have downplayed the plague entirely, stating Shipilova died of “pneumonia of unknown etiology”, finding “no micro-organisms linked to her professional activity”.

Inevitably, the incident has conjured up the spectre of bioweapons. Chinese and European accounts from the 1300s detail attempts by warring factions to weaponise the plague. Most infamous was the siege of Caffa in 1346, where Mongol attackers catapulted afflicted corpses into the city. More recently, the Soviet Union and imperial Japan experimented with weaponised plague during the second world war. But, evocative as this is, there is no evidence Shipilova’s death stemmed from anything so nefarious.

Whatever the cause of her tragic death, quarantining staff members and close contacts seems a reasonable caution rather than inherently ominous. Pneumonic plague is always concerning as it can spread from person to person by infected droplets. Untreated patients, however, get severely ill extremely quickly, drastically limiting transmission potential. While lab accidents happen, the likelihood of onward transmission remains vanishingly small. The chances of it triggering an epidemic are virtually zero, especially given infection is now readily treatable.

While we are still unclear as to what precisely transpired in Russia, this incident is a stark reminder that transparency is vital to mitigate the harms of such events. We can never afford to be complacent about our vulnerability to the microscopic.

DJI and Insta360 engaged in ‘civil war’ to become world’s top smart camera maker

The decline of action camera pioneer GoPro, which was rescued by a buyer last month, has allowed two Chinese rivals to rapidly dominate the market, as the country’s consumer tech brands take their cut-throat competition global.

DJI and Insta360, both based in the southern Chinese tech hub of Shenzhen, now account for more than 90 per cent of global handheld smart camera sales — a category that includes action cameras, panoramic view devices and gimbal cameras — according to consultancy IDC.

They are now engaged in a fierce battle to be the world’s top smart camera maker after displacing GoPro, which was the largest seller by revenue as recently as 2023, according to consulting firm Frost & Sullivan.

Their dominance shows how Chinese producers of consumer electronics are increasingly bringing their technology — and internecine competition — to markets such as the US and Europe.

Chris Pereira, founder of consultancy iMpact, said DJI and Insta360 were exporting Shenzhen’s “civil war”, referring to the intense competition between the city’s start-ups.

“A similar situation happened with robot vacuums,” said Pereira, noting how Chinese companies took over a market created by Roomba maker iRobot, which filed for bankruptcy last year.

“The strong competition . . . in the local Shenzhen market is generating extremely efficient . . . companies that are going overseas now.”

The performance of DJI and Insta360 contrasts with that of GoPro, which first commercialised action cameras in the early 2000s and was seen as a US tech champion.

Its sales have declined precipitously over the years as Chinese competitors innovated more quickly. By the second quarter of this year, GoPro snagged just 3 per cent of the market.

After years of struggling with debt, which totalled $92mn last month, it agreed to be taken over by a US optical hardware company for $285mn, far below its peak valuation of almost $12bn in 2014.

“In recent years, GoPro’s product iteration has been slow,” said Jacky Xue, a research manager at IDC. “Meanwhile, Chinese vendors such as Insta360 and DJI have seized the market opportunity. They all want to find new growth.”

In search of that growth, the two companies are increasingly encroaching on each other’s territory.

DJI, which dominates the global drone market, has aggressively expanded its offerings, unveiling a panoramic camera called Osmo 360 last year.

At the same time, Insta360, whose panoramic cameras are popular with extreme-sports enthusiasts and vloggers, launched a drone brand.

By the second quarter of this year, DJI held a 73 per cent share of the global smart camera market, while Insta360 took up 20 per cent, though it dominates niche categories such as panoramic cameras.

The pair have sued each other in the US, alleging patent infringements on a range of camera-related products.

DJI said it was investing in research on cameras, accessories and editing software. Recently, it has waded into new categories such as robot vacuums as its drone products face restrictions in the US market.

“Our success comes from our engineering and technical drive, along with a robust user feedback loop,” it said.

Insta360, which opened its first US flagship store in New York last month, told the FT it hoped to be the country’s top smart camera brand.

“We definitely want to become the market leader in this field,” said Max Richter, a vice-president at the company and former DJI staffer.

But it would have to be through rapidly introducing new products, he said, rather than competing on price, given DJI’s larger economies of scale.

The average price of a handheld smart camera has remained constant at about $380 since 2024, according to IDC.

“If we start to compete on price, we would just get into a spiral that probably ends badly for us but also for the whole market,” said Richter.

Tokenization of financial assets could be the biggest revolution on Wall Street since the move from paper to electronic trading. Don’t think Bitcoin. Think blockchain.

In Washington, the cryptocurrency industry just suffered a crushing legislative defeat. On Wall Street, the crypto-powered tokenization of financial markets is proceeding at a rapid pace. Despite the contradiction, the future is already here. Crypto technology is moving from the fringes toward the center of finance, whether or not legislators approve.

With the support of President Donald Trump, who made more than $1 billion from crypto ventures last year, Wall Street is embedding the blockchain technology powering Bitcoin into almost every corner of the financial system. Crypto and traditional finance firms are “tokenizing” or creating programmable versions of stocks, bonds, and other real-world assets and moving them onto blockchain networks that could one day replace the traditional venues that match trades, custody stocks, and settle transactions.

For investors, tokenized stocks could herald the beginning of round-the-clock trading, with instant purchases and access to the proceeds of sales. That would be particularly beneficial for overseas investors, who currently must confine their transactions to U.S. trading hours. Switching brokerages—a cumbersome process that takes days to weeks—also could happen almost instantly, while violations for using unsettled proceeds to trade stocks would largely disappear.

For the firms that power trades, the change potentially is even more profound. Much of Wall Street’s capital essentially is sitting today in the wrong place at the wrong time. Banks, brokerages, and asset managers position excess collateral at different institutions, wait for transactions to settle, and sometimes must borrow money because of asset-transfer delays. Modernizing the market’s plumbing could free up billions of dollars now idled as traditional systems process trades.

“This is a watershed moment,” says former New York Gov. Andrew Cuomo, who co-chairs OKXICE, a joint venture between New York Stock Exchange owner Intercontinental Exchange +0.54% and crypto platform OKX.

The JV announced on Oct. 4 that it plans to launch a tokenized market. Cuomo compares the move to tokenized trading to the birth of automated mortgage underwriting or the digitization of the stock market in the early 2000s. “I would hope that Wall Street as Wall Street still exists, but it will have to do business differently,” he says.

The emerging tokenization trend has little to do with Bitcoin, but relies on the blockchain technology that powers crypto trading. A blockchain is typically a decentralized digital database shared across a network of computers that is difficult to hack. Bitcoin traded on the first widely used blockchain, but companies are now applying the innovation to trade stocks and other assets. Even as the value of the crypto market has grown to nearly $3 trillion, the golden goose for some crypto firms has been winning the right to tokenize hundreds of trillions of dollars worth of traditional assets, including stocks.

Moving stock trades “onchain” could fundamentally change how a trade is processed. Transactions now move through a broker, an exchange, a clearinghouse, and other intermediaries, all of which carry separate ledgers that must be matched up before trades settle hours to days later. A blockchain-based system could remove these inefficiencies, creating instant settlement.

The benefits gained from blockchains could be enticing enough to eventually tokenize the entire trading system. In their base-case estimate, Citigroup researchers in June projected that the tokenized asset market will reach $5.5 trillion by 2030, up from $17 billion today. In addition to stocks, firms plan to bring bonds, funds, and even illiquid assets such as private equity and real estate onchain.

Several factors have coalesced to create this “liftoff” moment, including the rise of stablecoins, a type of token pegged to the dollar. Stablecoins have been used until now to trade cryptocurrencies and would be essential to executing tokenized asset trading in many models. Circle Internet Group +4.54% has issued one of the largest stablecoins, with a market capitalization of more than $73 billion, and various consortia of banks and retailers have announced in the past year that they are creating their own.

Some investors also see tokenized systems paving the way for artificial-intelligence “agents” to execute financial transactions. Agents work autonomously and can transact through traditional financial systems, but are much more valuable in a world where markets never stop trading.

The value of always-open markets became clear earlier this year when Trump launched major offensives against Iran outside normal market hours. With traditional futures markets closed, trading volume exploded on the crypto platform Hyperliquid in oil-related “perpetual” futures. By the time traditional markets opened, the crypto market had effectively priced in the news.

A major step forward in the tokenization of markets is slated to occur later this month, when the Depository Trust & Clearing Corp., or DTCC, which ensures that trades are processed and settled correctly, flips the switch on a platform allowing its customers to tokenize many stocks, exchange-traded funds, and other securities. DTCC’s platform will let participants convert tokenized assets to traditional assets, and vice versa, in part to avoid reducing liquidity. Tokenized stocks will transfer between holders almost instantly on a 24/7 basis.

“Honestly, for the past 10 years we’ve had innovation by press release,” says Nadine Chakar, global head of DTCC Digital Assets, alluding to small, pilot tokenization programs that never involved a critical mass of participants.

Not anymore. When DTCC ran a one-day trial of the new system this past summer, dozens of major financial institutions participated, including JPMorgan Chase +0.47%, Goldman Sachs Group, BlackRock, and the New York Stock Exchange. The goal of tokenized equity is to “move the markets to an always-on economy,” Chakar says.

Proponents of tokenizing stocks point to myriad advantages. The benefits of the DTCC platform mostly accrue to institutional investors, chiefly by speeding market plumbing. Under the platform that DTCC is launching this fall, investors’ trades will execute and settle through conventional methods, but DTCC participants can opt to receive a tokenized version of their shares. Institutions can then move tokenized shares onchain.

Speeding up such transfers could free up billions of dollars now held by institutions as excess collateral to deal with unexpected margin calls. Margin calls wouldn’t disappear, but institutions likely wouldn’t have to park extra collateral at multiple institutions simply because they don’t know where one might occur. Nasdaq estimates that the largest financial institutions collectively could earn up to $340 million a year in additional interest by reducing excess collateral and investing the money elsewhere.

Tokenization could also allow companies to pay dividends and execute stock splits programmatically, reducing the usual manual reconciliation work, which can take days. Nasdaq executives have said the shift could give international investors more access to U.S. stocks. The company plans to launch Nasdaq Equity Tokens, representing shares of publicly traded companies, next year.

“There have now been some use cases that have really unlocked the imagination of what tokenization can bring,” says Tal Cohen, Nasdaq’s president. “It’s no longer in the background.”

The Trump administration is trying to bring the benefits of tokenized stocks directly to retail investors. Trump was a vocal supporter of the Clarity Act, a sweeping crypto bill that, among other things, would have clarified that tokenized securities should be treated the same as traditional securities under the law. Senate Democrats blocked the bill for unrelated reasons, saying it didn’t do enough to rein in Trump’s personal crypto investments.

Yet, two days after the bill stalled, the Securities and Exchange Commission moved forward on its own, issuing an exemption that will allow crypto and traditional finance firms to proceed with trading venues that don’t need to comply with all the requirements governing exchanges. Traditional exchanges generally must execute trades at the best available price. They also can’t price shares in increments smaller than a penny, a rule designed to protect market liquidity. The new venues don’t have these requirements.

The goal “is allowing the market to evolve, monitoring its development, and using that insight to inform a nimbler and future-ready regulatory framework,” SEC Chairman Paul Atkins said in a statement.

One advantage of tokenization for retail investors would be the ability to switch brokerages readily. That would increase competition among firms, and potentially lower customers’ costs. An investor who wants to change brokerages now must typically initiate a transfer using the automated customer account transfer service, or ACATS, a process that can take several business days. A transfer would take minutes under a tokenized system, says Johann Kerbrat, a senior vice president at Robinhood Markets who helps manage the company’s crypto projects.

Not everyone is on board with the move toward tokenization. Tokenizing shares could make markets more prone to crashes by accelerating the pace at which investors can sell stocks, says Mark Hays, an associate director at Americans for Financial Reform, an advocacy group.

High-frequency trading, which accelerated in the late 1990s, contributed significantly to the frequency of so-called flash crashes, or sudden, steep selloffs, Hays says. Although 24/7 tokenized markets might bring convenience, there would also be a risk of overnight market crashes when liquidity is low. “If the blockchain is automatic and everything is happening in the wee hours of the morning, you need circuit breakers to keep things from crashing,” Hays says.

Until now, there has been little agreement about the structure a tokenized stock should have. Last year, before the SEC’s exemption, Robinhood launched tokenized versions of 200 U.S. stocks and funds for European traders. A more recent tokenized stock product launched by the company took the form of debt securities issued by Robinhood whose value adjusts based on the performance of the underlying equity. The tokens don’t give investors ownership or governance rights, or traditional dividends, although their prices are adjusted to account for dividend payments.

The potential confusion brought strong objections from Adam Aron, CEO of AMC Entertainment Holdings, whose shares were among those tokenized this way. Aron called the tokenization method “abhorrent” last month in a post on X and said that “it defeats the very ethos of share ownership.”

An AMC spokesman said Aron declined to comment further. A Robinhood spokesperson referred Barron’s to an article posted on X by CEO Vlad Tenev, in which he said that “a company should control the rights attached to its shares—not every lawful use of those shares once they’re in investors’ hands.”

The SEC exemption only allows the new trading venues to transact in tokenized shares that have the same rights as those issued conventionally. It also requires companies to give issuers such as AMC 30 calendar days to opt out of having their stock tokenized.

That requirement could become an important governor of how quickly some tokenized markets grow. On Oct. 4, the OKX and ICE joint venture became among the first to file notice that it is launching a tokenized market using the new SEC exemption. It said it initially plans to list 63 public companies, including Nvidia, Apple, Coca-Cola, and dozens of other recognizable names. The filing also disclosed that one company, Cerebras Systems, had already objected. Cerebras didn’t respond to requests for comment.

The other companies have at least several more weeks to opt out, depending on when they receive the notice. Cuomo says more objections are probably coming. “I’m sure many will opt out because it is very new and revolutionary,” he says.

Advertisement - Scroll to Continue

Another potential hurdle to adoption is the future of the SEC’s exemption, particularly if the next president is a Democrat who appoints an SEC chair more skeptical of crypto technology. The current exemption lasts five years, after which the agency can sunset or extend it, or replace it with new rules. Many industry players expect the SEC to undertake a formal rule-making process in the coming months. It would be harder to undo, but a Democratic-controlled commission could still effectively kill many tokenization projects.

Meanwhile, shares of some companies involved in the tokenization process could benefit directly if the tokenization trend continues. Companies with crypto trading platforms, such as Coinbase Global and Robinhood, have said for years that they are positioned to profit as their investors gain the ability to trade tokenized stocks in addition to cryptocurrencies.

Like Robinhood, Coinbase operates a tokenized equity program for investors outside the U.S., although its offshore tokens don’t have the same governance and dividend rights as traditional shares. Coinbase executives say they are still studying the SEC’s exemption but are eager to expand their tokenization efforts to U.S. investors. “Tokenization will represent the biggest upgrade to our financial system since we moved from paper to electronic trading,” says Coinbase Vice Chair Ryan VanGrack.

Bullish, a fintech company that operates a crypto exchange, said, days after the SEC announcement, that it is partnering with other companies to create standards for tokenized stocks, with an eye toward forming a coalition big enough to support a liquid tokenized stock market. The company announced in May that it was buying transfer agent Equiniti. Bullish CEO Tom Farley said on a recent earnings call that the SEC exemption “provides a mandate for us to go and talk to every one of our issuer customers about tokenizing their shares.”

Compass Point Research & Trading analyst Ed Engel upgraded Bullish to Buy in September, saying the company “is among the few public companies that can monetize this trend.”

Securitize, which went public through a special purpose acquisition company, or SPAC, merger in July, is another company focused on tokenization. Securitize launched a program on Oct. 8 that will allow the trading of tokenized shares of a dozen major companies, including Apple, Microsoft, and Nvidia. “The opportunity to bring securities onchain is significant,” said Securitize CEO Carlos Domingo, in a statement to Barron’s.

One of Securitize’s advantages is the SEC’s decision to require tokenized shares to carry the same rights as traditional ones, says StoneX Group analyst Mark Palmer.

The SEC order’s “fine print reads like a blueprint of the model for tokenized equities that [Securitize] has spent years building,” Palmer wrote in a research note.

Right now, the stock market’s chief regulator is paving the way for a tokenized future—one that establishes blockchain technology’s commercial purpose. If current tokenization initiatives succeed without harm to companies or investors, calls to overhaul Wall Street’s plumbing are bound to grow.

Yes, the market is being led by a handful of highfliers. But that isn’t a bad thing.

A yawning gap has opened up inside the stock market that to some looks like an abyss. We see an opportunity.

On the surface, the stock market is doing just fine. The S&P 500 index ended September with a loss of just 0.5%, and is less than 1% below the record high it set in August. That’s certainly better than might be expected, given that the Federal Reserve has started raising interest rates, the 10-year Treasury yield just hit its highest level since 2002, and the Iran war continues with no end in sight, giving a boost to energy prices. Meanwhile, there are still lingering questions about the artificial-intelligence trade, from the fundamental—how will Anthropic and OpenAI turn a profit?—to the existential—will artificial intelligence mean the end of humankind?

A glimpse beneath the surface reveals that these concerns are affecting stocks—just not at the index level. While the market-cap-weighted S&P 500 fell 0.5% in September, an equal-weighted version dropped 5% and is on pace for its seventh straight weekly loss, a losing streak previously seen only in 2002 and 2022.

It gets worse. Not since 2000 have more S&P 500 stocks made 52-week lows than 52-week highs when the overall index was within 2% of an all-time high, according to BTIG Chief Market Technician Jonathan Krinsky, an ominous sign for a market that has drawn more than a few comparisons to the dot-com bubble.

The divide in the market is wide and getting wider. Out of the 11 S&P 500 sectors, only information technology and communication services rose in September. Chips, though still down from their record high, contributed, too, with the PHLX Semiconductor Index surging 9.5% in September, while the Roundhill Magnificent Seven exchange-traded fund rose 4.3%. And it’s impossible to overstate the impact of Meta Platforms, whose 27% September Muse-inspired rally added 0.6% to the S&P 500’s monthly performance.

Some worry that this strange dynamic presages a broader decline for the S&P 500. Krinsky acknowledges that beaten-down sectors could bounce, but says the most likely scenario is that “the leaders succumb and major indexes catch down to what the majority of stocks have been doing.” Famed bond investor Jeffrey Gundlach describes the problem in arboreal terms, telling David Rosenberg that, like a particular century-old silver maple in his yard, the market has hollowed out after rotting from the inside and is now “on the edge of complete failure.”

The market’s bad breadth, however, looks quite different from other episodes of narrow markets. Investors aren’t simply bidding up a small set of winners into nosebleed territory. Nvidia, the world’s biggest stock, has seen its price/earnings ratio fall below the overall market’s even as it has rallied 24% this year. Meta’s valuation has fallen this year, as well, despite its September surge. That’s in contrast to the dot-com bubble, when the valuations of heavyweights like Microsoft and Cisco Systems surged.

The poor breadth also looks more rational this time around. Rather than the bifurcation between the “old” and “new” economies seen during the dot-com bubble, investors are trying to gauge the winners and losers of AI in real time. Take the Meta rally. As that stock soared, shares of companies like Planet Fitness and New York Times stumbled, hit by fears that the new personal AI agent, Muse, will help paying customers cancel subscriptions to services that they no longer use. Charles Schwab and Bank of America, meanwhile, dropped over concerns that the technology would flush investor cash out of ultralow-yielding accounts. Investors aren’t gripped by some speculative frenzy; they’re articulating a holistic and reasonable thesis about the winners and losers in an AI-driven economy.

A similar dynamic is playing out elsewhere in the market. The surge in the price of diesel has benefited refiners like Valero Energy and Phillips 66, which hit fresh all-time highs last month, while air carriers like JetBlue Airways and American Airlines Group have been struggling. And while rising rates are generally negative for equities, a higher-yield environment is certainly harder on companies that have large amounts of floating-rate debt, or whose potential cash flows are many years away.

“Maybe the real theme,” says John Marshall of Carrick Lane, “is that the trends that are most prominent in the market now all have winners and losers.”

Perhaps the focus on market breadth, then, is missing the point. Owen Lamont, portfolio manager and researcher at Acadian Asset Management, says it’s “crazy” to argue that the stock market is risky just because it’s highly concentrated. “If you took the Magnificent Seven, and broke each into seven smaller firms so it was the Magnificent 49, why would that be any better?”

The historical record suggests that breadth isn’t the bad omen that you might think it is. “It’s always preferable to see a market rally on broad participation rather than a small group of stocks, and the durability of narrow rallies is more dependent on that narrow group of stocks,” writes Bespoke Investment Group. “Contrary to popular wisdom, though, weak breadth alone hasn’t been a reliable bearish short-term signal. In prior periods when breadth deteriorated but price held up relatively well, forward returns saw little in the way of a negative impact.”

After all, indexes, unlike silver maples, are recomposable. As failing companies lose value, they are replaced within the index by up-and-coming names—not because the index masters are betting that these incoming companies will outperform the outgoing ones, but simply because they’re striving to maintain an index that continually gives the most accurate read on the segment of the market in question, such as large-cap U.S. stocks in the case of the S&P 500. Out of the 500 companies originally included in the S&P 500 in 1957, only about 50 remain. Like Theseus’ ship, the market keeps sailing on.

And the weather might soon get more temperate. Like Krinsky, Morgan Stanley Chief U.S. Equity Strategist Mike Wilson believes that “the divergence between the index and breadth must be reconciled,” and he says the most likely scenario is that the overall market will slip in the weeks ahead. But he remains fundamentally bullish. “If bond volatility doesn’t calm down, we see breadth and price meeting in the middle in the next month followed by a strong finish to the year,” Wilson writes. “If bond volatility subsides sooner, breadth is likely to catch up to index price, driving both higher in the near term.”

One thing that could calm the bond market is a resolution to the conflict in Iran. “I’m in the camp that we’re going to find a solution in Iran,” says KKM Financial CEO Jeff Kilburg. “I think we get out this month, and then we see crude oil crack, and we see rates crack,” with the 10-year yield falling back below 5%.

Another catalyst: the start of earnings season. It’s a truism on Wall Street that investors seem to worry about all of the macro headwinds during the months between earnings seasons, only to be reminded just how good they have it when the next crop of reports comes around. And third-quarter earnings season should be better than good. Analysts collectively estimate that S&P 500 companies will report year-over-year earnings growth of 29% for the quarter, which would mark the third-straight quarter of earnings growth about 25%, according to FactSet.

A cooling-off in oil prices and rates, “in conjunction with earnings season showing us yet more 25%-plus profit growth, is the catalyst for a run-up to 8000 in the S&P,” Kilburg says, referring to a level 4.4% above Thursday’s close.

Finally, the overhang of the midterm elections will pass after Nov. 3. “Historically, U.S. equity volatility has tended to rise in the months leading into the midterm elections,” writes Dubravko Lakos-Bujas, head of global markets strategy at J.P. Morgan. Given that implied volatility has remained remarkably low, he does think there’s room for the Cboe Volatility Index, or VIX, to rise from here. But he adds that the elections “should serve as an important market-clearing event. Historical performance in the months following the midterms tends to be positive.”

The great hope is that all of the major drivers of macro uncertainty will recede, while companies continue to report boffo earnings. That may be a bit too much to expect. But we do know that overall stock market valuations have fallen meaningfully this year, and have continued to decline during the past eight weeks of market churn.

Far from being complacent, investors are acutely aware of the pressures caused by rising rates, which is why the average stock has performed so poorly. In fact, according to Goldman Sachs, overall U.S. equity sentiment has fallen sharply into the negative, and is now as low as it was in March. As long as the AI narrative holds together, and earnings aren’t terribly disappointing, the market is setting up for strong close to 2026.

Even if the rising tide doesn’t lift every stock.