Slowing an investment boom on the scale of the AI build-out is a tricky business — as yesterday's rout in the stock market demonstrated.

Why it matters: Americans may have voted in the midterms to tap the brakes on the AI race, but curbing the investment boom — or some might say, gently deflating the AI bubble — is easier said than done.

The big picture: The historical record shows that once major investment booms are in motion, it can be difficult to temper them without sparking a stampede of investors out of assets.

• That's why market collapses have featured prominently in past investment booms like those in canals, railroads and the internet.
• "We can't control it," John D. Turner, a finance professor at Queen's University Belfast who co-wrote "Boom and Bust: A Global History of Financial Bubbles," tells Axios.
• "People just get spooked, and as soon as they do, then, you know, prices head down. And that's what happens in historical bubbles, and it can happen quite quickly."

State of play: Tech stocks plunged yesterday as the week has developed into a major test for the market.

• The Nasdaq Composite Index fell 5.5% — its worst daily showing since the quasi-panic of President Trump's "Liberation Day" tariffs in April 2025 — and is down roughly 10% from the high it hit a month ago.
• On Wall Street, a drop of that magnitude is known as a "correction," a term of art indicating a sell-off more serious than a garden-variety decline.
• The Nasdaq 100, a measure of the largest tech stocks, is now down roughly 14% since its Oct. 7 peak. Its 6.4% drop yesterday was its steepest since the March 2020 pandemic panic.
• The SOX semiconductor index also endured its worst day since the pandemic, plummeting 12.3% yesterday. The broader S&P 500 index slumped 4.4%.

Reality check: The 4-year-old bull market for the blue chips still holds: The S&P 500 is off 6% from its high and is up 7.4% on the year.

Catch up quick: The overwhelming rejection of AI in this week's elections — from wins for AI-cautious Democrats in governors' races in hotbeds of development to overwhelming support for ballot measures restricting data center construction — has pushed investors to reevaluate their previous bets on companies profiting from the data center boom.

Zoom in: Many of the proposed constraints on AI are limited in nature and do not apply to projects already underway. Even so, the market reaction has been ferocious.

• Stocks tied to the AI buildout were hit particularly hard yesterday, with makers of everything from fiber optic connections (Coherent down 13%, Ciena down 15%) to HVAC systems (Comfort Systems, -11.6%), servers (Dell dove 15%) and semiconductors (Advanced Micro Devices, -17.3%) knocked down, some after enduring deep losses Wednesday.
• Engineering, electrical and industrial equipment stocks closely tied to the boom also collapsed. Quanta Services and Vertiv Holdings plunged 12% and 14%, respectively.
• The so-called hyperscalers — Amazon, Alphabet, Microsoft, Meta and Oracle — were all down, too, though their losses were less lurid, as they could benefit from any slowdown in AI spending as a result of increased regulatory scrutiny.

Between the lines: To some, the sudden tumble in the markets — futures were actually up a bit in premarket trading — suggested that a large institutional investor may have moved to sell a sizable chunk of a portfolio that had turned south after Wednesday's post-election slump in AI shares.

• Elsewhere, retail investors, who in past market downturns have swarmed in to buy the dip, instead dumped shares, acting as heavy sellers of tech stocks yesterday, said Jai Malhi, senior equity strategist at Vanda Research, which tracks trading by individual investors.

What we're watching: For any indications that the volatility hitting the AI trade could prompt Anthropic and rival OpenAI to reconsider plans to offer shares to the public.

Refrigerator hacks. $6.3 billion crypto heists. 14-country espionage campaigns carried out in just 11 days. All in the span of a week.

Why it matters: State-backed hackers are finding new ways to launch cyberattacks at remarkable speed and scale, with AI and automation threatening to widen the gap between attackers and defenders.

Driving the news: Hackers tampered with refrigeration systems at dozens of European grocery stores in one night earlier this week — raising temperatures and spoiling food in a sweeping attack some linked to Russia.

• "It's hard to imagine this is not the work of a nation state pretending to be a criminal actor," Cynthia Kaiser, former deputy director of the FBI's cyber division, told Axios. "Given the concentration around Europe, Russia is a reasonable first suspect."
• French blood banks were also affected in the attacks on European refrigeration systems, authorities said Wednesday.
• SentinelOne researchers reported Tuesday that Russian-aligned hackers have previously claimed attacks on European refrigeration systems.
• Meanwhile, the Pentagon has also been investigating a series of suspected hacks of refrigeration systems on its military bases over the summer. Nextgov reported that investigators suspect Iran was behind those attacks. Others have suggested potential Iran ties in Europe as well.

The big picture: No clear evidence has yet linked the attack on food supply through smart refrigeration to AI or automation, but nation-state hackers have used those tools in other breaches to speed up their operations.

• Google researchers said last week that multiple hacking groups have used automated security testing tools in recent attacks. One China-linked group gained deep access to a victim's network in as little as 15 minutes.
• North Korean hackers behind last week's record-setting Binance heist also likely used automation to move stolen cryptocurrency and cover their tracks, according to Chainalysis.

Case in point: China-linked hackers used an AI-powered tool called Artex to target more than 200 organizations across 14 countries in just 11 days last month, CrowdStrike found.

• The group, known as Vapor Panda, targeted government agencies across Latin America, the Middle East and Europe, along with U.S. organizations and telecommunications providers in several countries.
• The hackers paired Artex with DeepSeek and a commercial AI model to run agents that automated much of the operation.

Between the lines: Artex is one of several AI-powered tools automating tasks once handled by skilled human hackers.

• Adam Meyers, CrowdStrike's senior vice president of counter adversary operations, told Axios he tested Artex himself and found it could probe websites for weaknesses far faster than a human.
• "It was doing the stuff that you, as a pen tester, would do," Meyers said. "But it does it really at a scale and a speed that I think a human would not be able to do."
• Other AI-powered security testing tools, including Cairn and Strix, could be repurposed for similar attacks, Meyers said.

• Anthropic may have doubled the pace at which it is developing AI models, according to new research from METR, a research organization that's part of an experiment in "embedded" evaluations.
• Dario Amodei gets the New York Times treatment, which looked into how his "thinking on artificial intelligence was shaped."

Ina here, letting you know that an American has set a new Guinness world record by going more than 80 mph while sand skiing. Also, there's a thing called sand skiing.