U.S. stocks had their worst day since 2022 on Wednesday amid a broad pullback in tech companies as Wall Street traders sought to reduce their exposure to firms that have made big bets on artificial intelligence.

The tech-heavy Nasdaq index closed down 3.6%, while the broader S&P 500 index closed down 2.3% — both their worst performances in more than 18 months. The Dow Jones Industrial Average fell 1.25%.

The rout was led by Tesla, whose shares fell 12.3% for its worst day since 2020, and Google parent Alphabet, which fell more than 5% for its worst day since January.

Tesla reported Tuesday afternoon that its auto revenues fell 7% compared with the previous quarter, and CEO Elon Musk said in a follow-up earnings call that the company’s planned robotaxi rollout would be pushed back.

Although Alphabet reported earnings Tuesday that were in line with analysts’ expectations, traders appeared to seize on remarks CEO Sundar Pichai made on the company’s earnings call that signaled the tech world’s booming investments in artificial intelligence were not going to pay off in a short time frame.

‘I think we are in this phase where we have to deeply work and make sure on these use cases [for AI products], on these workflows, we are driving deeper progress on unlocking value, which I’m very bullish will happen,’ Pichai said. ‘But these things take time.’

Steve Sosnick, chief strategist at Interactive Brokers financial group, said Wall Street took that as a signal to sell off shares that had enjoyed the frenzied growth that tech stocks have been experiencing in recent months.

‘We’re seeing some nervous profit-taking in some of the stocks that have been leveraged to AI that a lot of investors have come to rely on as a consistent source of stock market gains,’ Sosnick told NBC News.

Alphabet also reported weaker-than-expected ad revenue from YouTube, which Google has owned since 2006.

Other major tech names having major losses Wednesday included Nvidia, the computer chip maker powering much of the AI revolution, whose shares fell more than 6% for their worst day since 2022; Facebook parent Meta was down 5%; Microsoft fell 3.5%; and Amazon lost 3%.

The major indices have been on a relatively consistent and positive run. Even after Wednesday’s dip, the S&P 500 remains up 13.8% in 2024, with the Nasdaq up 15.5% and the Dow up 5.7% in that time.

Wednesday’s sell-off comes amid renewed expectations for an interest-rate cut from the Federal Reserve in response to a slowing economy. While traders now say the Fed’s first cut of the post-pandemic period is virtually guaranteed by September, former Federal Reserve Bank of New York President Bill Dudley wrote Wednesday that the Fed needs to strongly consider announcing a cut at its meeting next Wednesday.

‘Although it might already be too late to fend off a recession by cutting rates, dawdling now unnecessarily increases the risk,’ Dudley, now an executive at UBS financial group, wrote in a column for Bloomberg News.

Sosnick said Wednesday’s stock sell-off was not a total referendum on the broader state of the economy. Year to date, the S&P 500 has still had healthy gains of about 15%, while the Nasdaq is up about 18% and the Dow Jones Industrial Average is up 6%.

‘This is much more about a little bit of vertigo in names that have climbed a lot this year,’ Sosnick said.

But signs of a broader economic pullback continue to mount: The U.S. unemployment rate is rising, excess savings from the pandemic have been exhausted, and consumer borrowing stress is at fresh highs.

‘We expect relatively weak economic growth in the second half of 2024 and early 2025,’ Ian Shepherdson, chief economist at Pantheon Macroeconomics research group, wrote in a note to clients Wednesday.

At Berkshire Hathaway’s annual investor meeting earlier this year, Warren Buffett and his top insurance executive Ajit Jain issued a headline-grabbing warning that Berkshire would exercise caution regarding cyber insurance — in fact, it advised insurance agents to only sell cyber policies if they absolutely had to do so to satisfy a client, and to expect losses.

A primary reason cited is the difficulty in assessing the scale of losses possible from a single occurrence that spreads across technology systems, with Jain giving the hypothetical example of when a primary cloud provider’s platform “comes to a standstill.”

“That aggregation potential can be huge, and not being able to have a worst-case gap on it is what scares us,” he said.

Jain’s hypothetical seemed prescient when a quality control issue from cybersecurity firm CrowdStrike caused a worldwide IT outage that halted flights and freight, shuttered retail outlets, and caused hospitals to resort to charting on paper.

“Insurers have been worried about something like what happened with CrowdStrike since cloud adoption happened,” said Dale Gonzales, chief innovation officer at Axio, a cyber security risk analysis company.

But Gerald Glombicki, a senior director in Fitch Rating’s U.S. insurance group, believes the cyber insurance industry largely priced in the CrowdStrike meltdown correctly, and he expects it to be manageable rather than catastrophic for the cybersecurity insurance firms..

“It will have an impact because there will be losses,” said Glombicki, “but the modeling largely got it right. Mostly, we think the industry will handle it OK. There might be some issuers that mispriced policies,” he added. 

Fitch estimates that the number of insured losses will not exceed $10 billion, ending somewhere in the mid- to high-single billions and that the industry largely priced those in.

The cybersecurity insurance market did get lucky, in some respects, with the CrowdStrike meltdown. For one, there were no significant physical damages, such as explosions at power plants, dams bursting, or fires caused by overheating equipment, which are becoming a bigger cyberterrorism risk.

“Cyber events that have more of a physical consequence would be much bigger in size or scope in terms of losses,” Glombicki said.

Additionally, even though CrowdStrike is widely deployed, its market share, estimated at 17% by Fitch, is large but limited in total impact. Among the companies that did use CrowdStrike, the worst impacted seemed to be on businesses that need 24/7 availability, like hospitals and airlines, Glombicki said.

Another factor in holding down losses and distributing them unevenly across the globe is that the CrowdStrike failure impacted places like Australia and Pacific Asia in the middle of the business day, but other markets, including the U.S., were hit during the night or early morning and many businesses were able to get systems back up within hours.

Not all cyber experts are expressing as much confidence at this point. Josephine Wolff, an associate professor of cybersecurity policy at Tuft University’s Fletcher School who has been studying the evolving market for the past several years, suspects the CrowdStrike meltdown will send shock waves through the nascent cyber insurance market.

“It’s still pretty early to assess the volume of claims that insurers are going to see due to CrowdStrike, but I sense that there will be a lot of business interruption claims across all industry sectors, just based on the impacts we’ve seen covered in the news, and that it will be a very bad situation for insurers,” Wolff said.  

Wolff says the duration of the outages will influence the claims. Some businesses were out for hours; others were still struggling days later.

She compared it to the NotPetya cyberattacks launched by Russia in 2022, which halted much of the world’s freight.

“It’s possible that since some of these outages were shorter than what we saw after NotPetya, the claims may be smaller, at least in some cases,” Wolff said. However, she points out that the CrowdStrike glitch significantly impacted businesses, which was not the case with NotPetya.

“The U.S. is far and away the region with the highest rates of cyber insurance adoption, so I am guessing that this will be a bigger event for the cyber insurance industry both in terms of how many claims are filed and how big they are,” Wolff said.

In addition to unequal impact, cyber insurance policies themselves vary widely.

“Cyber insurance policies can be dramatically different. There is no standardization; terms and conditions can differ within a company depending on who wrote the policy,” Glombicki said.

Insurers are already cognizant of the unique challenges that cybersecurity poses for them, Gonzales said. As a result, the companies try to spread losses smartly by diversifying what is covered. However, the problem with cyberspace and ensuring its security is that it is still relatively unknown. But he doesn’t think it will drag down the whole insurance market.

“The losses won’t be as bad as hurricanes last year,” Gonzales said, adding that the comparison isn’t quite apples to apples since far more entities are insured in hurricane zones than there are cyber insurance policies. 

Gonzales says the primary claims will be for business interruption, which some policies specifically exclude anyway. But he does predict the CrowdStrike incident will cause litigation.

“CrowdStrike will be sued. There will be litigation,” he said.

“Everyone exceedingly well understands fire insurance because it has been litigated to death,” Gonzales said. 

Cyber insurance, on the other hand, hasn’t yet been litigated enough to establish protocols and precedents.

“The litigation will help define business interruption and define third-party culpability. The industry could use some defining, and hopefully, litigation fixes it,” Gonzales said. “Cyber events are evolving in ways that are slightly unpredictable. It creates a very dynamic environment,” he said, but he added, “I don’t think the CrowdStrike event will drastically change how people think about insurance.”

Ironically, the Crowdstrike event could create more interest in cybersecurity and draw more customers into the market, Glombicki said. “Boards will be asking about it,” he said.

Wall Street’s favorite recession signal started flashing red in 2022 and hasn’t stopped — and thus far has been wrong every step of the way.

The yield on the 10-year Treasury note has been lower than most of its shorter-dated counterparts since that time — a phenomenon known as an inverted yield curve which has preceded nearly every recession going back to the 1950s.

However, while conventional thinking holds that a downturn is supposed to occur within a year, or at most two years, of an inverted curve, not only did one not occur but there’s also nary a red number in sight for U.S. economic growth.

The situation has many on Wall Street scratching their heads about why the inverted curve — both a signal and, in some respects, a cause of recessions — has been so wrong this time, and whether it’s a continuing sign of economic danger.

“So far, yeah, it’s been a bald-faced liar,” Mark Zandi, chief economist at Moody’s Analytics, said half-jokingly. “It’s the first time it’s inverted and a recession didn’t follow. But having said that, I don’t think we can feel very comfortable with the continued inversion. It’s been wrong so far, but that doesn’t mean it’s going to be wrong forever.”

Depending on which duration point you think is most relevant, the curve has been inverted either since July 2022, as gauged against the 2-year yield, or October of the same year, as measured against the 3-month note. Some even prefer to use the federal funds rate, which banks charge each other for overnight lending. That would take the inversion to November 2022.

Whichever point you pick, a recession should have arrived by now. The inversion had been wrong only once, in the mid-1960s, and has foretold every retrenchment since.

According to the New York Federal Reserve, which uses the 10-year/3-month curve, a recession should happen about 12 months later. In fact, the central bank still assigns about a 56% probability of a recession by June 2025 as indicated by the current gap.

“It’s been such a long time, you have to start to wonder about its usefulness,” said Joseph LaVorgna, chief economist SMBC Nikko Securities. “I just don’t see how a curve can be this wrong for this long. I’m leaning toward it being broken, but I haven’t fully capitulated yet.”

Making the situation even more complicated is that the yield curve isn’t the only indicator showing reason for caution about how long the post-Covid recovery can last.

Gross domestic product, a tally of all the goods and services produced across the sprawling U.S. economy, has averaged about 2.7% annualized real quarterly growth since the third quarter of 2022, a fairly robust pace well above what is considered trend gains of around 2%.

Prior to that, GDP was negative for two straight quarters, meeting a technical definition though few expect the National Bureau of Economic Research to declare an official recession.

The Commerce Department on Thursday is expected to report that GDP accelerated 2.1% in the second quarter of 2024.

However, economists have been watching several negative trends.

The so-called Sahm Rule, a fail-safe gauge that posits that recessions happen when the unemployment rate averaged across three months is half a percentage point higher than its 12-month low, is close to being triggered. On top of that, money supply has been on a steady downward trajectory since peaking in April 2022, and the Conference Board’s index of leading economic indicators has long been negative, suggesting substantial headwinds to growth.

“So many of these measures are being questioned,” said Quincy Krosby, chief global strategist at LPL Financial. “At some point, we’re going to be in recession.”

Yet no recession has appeared on the horizon.

“We’ve got a number of different indicators that just haven’t panned out,” said Jim Paulsen, a veteran economist and strategist who has worked at Wells Fargo among other firms. “We’ve had a number of things that were recession-like.”

Paulsen, who now writes a Substack blog called Paulsen Perspectives, points out some anomalous occurrences over the past few years that could account for the disparities.

For one, he and others note that the economy actually experienced that technical recession prior to the inversion. For another, he cites the unusual behavior by the Federal Reserve during the current cycle.

Faced with runaway inflation at its highest rate in more than 40 years, the Fed started raising rates gradually in March 2022, then much more aggressively by the middle part of that year — after the inflation peak of June 2022. That’s counter to the way central banks have operated in the past. Historically, the Fed has raised rates early in the inflation cycle then started cutting later.

“They waited until inflation peaked, and then they tightened all the way down. So the Fed’s been completely out of synch,” Paulsen said.

But the rate dynamics have helped companies escape what usually happens in an inverted curve.

One reason why inverted curves can contribute to a recession as well as signal that one is occurring is that they make shorter-term money more expensive. That’s hard on banks, for instance, that borrow short and lend long. With an inverted curve hitting their net interest margins, banks may opt to lend less, causing a pullback in consumer spending that can lead to recession.

But companies this time around were able to lock in at low long-term rates before the central bank starting hiking, providing a buffer against the higher short-term rates.

However, the trend raises the stakes for the Fed, as much of that financing is about to come due.

Companies needing to roll over their debt could face a much harder time if the prevailing high rates stay in effect. This could provide something of a self-fulfilling prophecy for the yield curve. The Fed has been on hold for a year, with its benchmark rate at a 23-year high.

“So it could very well be the case that the curve’s been lying to us up until now. But it could decide to start telling the truth here pretty soon,” said Zandi, the Moody’s economist. “It makes me really uncomfortable that the curve is inverted. This is one more reason why the Fed should be lowering interest rates. They’re taking a chance here.”

Renewable energy demand will triple over the next seven years as data center growth accelerates to facilitate the proliferation of artificial intelligence, NextEra Energy CEO John Ketchum said Wednesday.

NextEra added 3,000 megawatts of renewable and storage projects to its order backlog in the second quarter. Of those, 860 megawatts — or 28% — come from agreements with Google to power the tech company’s data centers.

“This marks our second best origination quarter ever,” Ketchum told analysts on the company’s earnings call Wednesday. “These results support our belief that the bulk of the growth demand will be met by a combination of renewables and battery storage.”

NextEra’s business with tech and data center customers currently stands at seven gigawatts of renewable assets in operation and in backlog, said Brian Bolster, NextEra’s chief financial officer.

NextEra stock was up 3.5% in early afternoon trading. It is the largest power company in the S&P utilities sector by market capitalization and operates the largest renewable portfolio in the U.S.

Shares have gained 24% year to date and 12% over the last three months, as investor enthusiasm over the company’s position to meet growing U.S. power demand.

NextEra expects power demand to grow four times faster over the next decades compared to the prior 20 years on demand from data center, manufacturing and the electrification of the economy, Ketchum said.

Consulting firm Rystad Energy recently forecast that data centers and the adoption of electric vehicles alone will result in additional 290 terawatt hours of electricity demand in the U.S. by 2030. That’s equivalent to the entire power demand of Turkey, according to Rystad.

Executives at some of the biggest utilities in the U.S. have warned that failure to meet this demand will jeopardize the nation’s economic growth. Rebecca Kujawa, CEO of NextEra Energy Resources, a subsidiary NextEra Energy, said it will take time to nail down concrete numbers on exactly how much demand is coming from data centers in particular.

“But there is no escaping the fact that these are very large numbers and numbers that I don’t think any utility across the industry has seen before,” Kujawa said Wednesday. “From a practical standpoint, it’s going to take a couple of years for this really to materialize and utilities to be able to absorb it and serve it.”

Natural gas is also expected to play a key role in meeting power demand, though there is an ongoing debate about how the power mix will break down between gas and renewables. Producers and pipeline operators have argued that renewables, which are dependent on sun and wind conditions, will need gas as backup to ensure reliable power.

Alan Armstrong, CEO of pipeline operator Williams Companies, told CNBC last week the U.S. risks falling behind in the AI race if it doesn’t embrace natural gas as a power source.

Ketchum said natural gas has an important role to play as a bridge fuel during the energy transition. NextEra owns and operates a natural gas fleet in Florida. But the CEO said renewables come at a lower cost and are faster to deploy.

Building new natural gas generation is “more expensive in most states, is subject to fuel price volatility, and takes considerable time to deploy given the need to get gas delivered to the generating unit and the three- to four-year waiting period for gas turbines,” Ketchum said.

With power demand expected to surge, there is growing interest in nuclear energy as a source of reliable, carbon free energy. Ketchum indicated Wednesday that NextEra is considering restarting the Duane Arnold nuclear plant in Palo, Iowa, though it would require a thorough assessment. The plant ceased operations in 2020.

“We would only do it if we could do it in a way that is is essentially risk free with plenty of mitigants around the approach,” Ketchum said Wednesday. “There are a few things that we would have to work through but yes — we are we are looking at it.”

NextEra is rated as a buy equivalent by 70% of Wall Street analysts, with an average price target of $79.12 per share, suggesting nearly 10% upside from Tuesday’s close of $72.11.