TheChek News Flow

Thames Barrier upgrade could cost at least £20bn, senior government officials warn

The Thames Barrier that protects Londoners from flooding may need to be replaced up to two decades earlier than anticipated and could cost at least £20bn, according to senior government officials who warn that an increase in extreme weather has put the vital defence system under strain.

Its 10 steel-capped gates that can restrict the flow of water were erected in the 1980s and built to last until 2030, but the public body that runs and maintains the system has since said the infrastructure can operate until 2070.

It provides crucial protection for the Houses of Parliament, the O2 Arena and Tower Bridge, as well as 116 railway and Tube stations, 300km of major roads and £321bn worth of residential homes. 

The Environment Agency (EA) had been planning to make a decision by 2040 on how to bolster the barrier, which spans the River Thames at Woolwich Reach in east London.

But senior government officials at the agency told the FT they believe the barrier might need to be replaced between 2050 and 2060, given its gates have been used more than anticipated owing to an increase in extreme weather caused by climate change. This has led to a degradation of the infrastructure, but also prevented maintenance from being carried out.

Existing climate projections indicate a new barrier will be needed from 2070.

The officials said a decision will probably need to be made imminently about how the barrier should be upgraded as it could take up to 30 years to erect a replacement. They estimate the cost of the works will be close to £20bn, more than double previous estimates of between £6bn and £9bn.

“£6bn is the figure announced publicly for a new Thames Barrier, which is ridiculously low,” one government figure said. 

“Whenever you’re working in a live waterway, especially an estuary like the Thames which has huge fluctuations each day, it’s just mega-expensive work,” they added, noting that even £20bn was a conservative estimate.

The current barrier was decades in the planning and took eight years to build at a cost of roughly £535mn in 1982, the equivalent of about £2.6bn today. 

“We totally forget how important the Thames Barrier is. We would be two metres underwater in Westminster if it wasn’t there,” said one senior government official. “Climate change hasn’t even really started yet. What it will mean in the UK is cold, wet winters and very intense summer rainfall.”

Options for a replacement include building a new barrier with locks on the existing site to improve resilience or creating a structure further downstream. Another possibility is installing “flood storage areas” — reservoirs and marshlands — to which surplus water can be diverted.

The EA has announced plans to fortify defences upstream, between Charlton in south-east London and Teddington in the city’s south-west, by 2050 to reduce reliance on the barrier itself.

However, officials say there is scepticism within the department about the feasibility of some of these plans, given they would involve engaging with thousands of different landowners, most of whom could be asked to front the costs themselves.

The barrier was also designed to defend against the UK’s north-south tilt, a geographical process by which the south-east of England is gradually sinking.

Since it was built, the barrier’s gates have been raised more than 221 times, including 50 times in 2013-14 — the busiest ever year owing to storm surges that pushed up the tides.

The number of gate closures is increasing, with the annual total expected to reach 50 a year by 2035.

Experts say the barrier has been so successful that Londoners have largely forgotten how vulnerable the city once was to flooding. In the 1970s and 1980s, sirens warned residents when floodwaters were rising, while posters and television adverts offered instructions on what to do. 

Caroline Douglass, executive director for flood and coastal risk management at the EA, said: “Flooding has a devastating impact on communities and will only worsen as our climate changes.

“We are actively assessing options for managing tidal flood risk, including consideration of options for upgrading or replacing the Thames Barrier and other associated barriers.”

She added that the EA is “exploring options for funding the essential infrastructure upgrades”.

Databricks is back raising another $5 billion, after it raised that amount eight months ago. The largest fundings also went to an AI neolab, data center and electricity storage, defense, coding and biotech. Let’s take a look.

1. Databricks, $5B, data platform: Databricks has surpassed a $7 billion revenue run rate, with more than 80% year-over-year growth in Q2. The San Francisco-based company raised $ 5 billion in a funding round led by Coatue, with participation from Blackstone, MGX, T. Rowe Price, and new investor Sixth Street Growth, and raised its valuation to $190 billion. Databricks was valued at $134 billion back in December 2025 when the company announced it had surpassed a $4.8 billion revenue run rate. From its Series A funding of $13 million in 2013, the 13-year-old company has raised around $25 billion in funding over all time, per Crunchbase data.

2. River AI, $1.1B, AI reinforcement learning: Founded earlier this year, River AI raised $1.1 billion across its seed and Series A rounds led by AMP PBC and General Catalyst, with strategic investment from Nvidia and AMD Ventures. Its founder, Igor Babuschkin, has been at the center of AI developments at Google DeepMind, OpenAI and xAI over the past decade. The Palo Alto, California-based company seeks to build AI that is personalized and trained directly on what a company or person needs to accomplish, giving the end user control.

3. Form Energy, $750M, electric grid: Form Energy raised a Series G funding round led by T. Rowe Price. The Massachusetts-based company builds longer-lasting 100-hour batteries for grid electricity storage. A host of investors participated, including private equity, venture and angels, and infrastructure and climate-focused investors Blindspot Ventures, Breakthrough Energy Ventures, Energy Impact Partners, GE Vernova, Gigascale Capital, NGP Energy Technology Partners, Prelude Ventures and the TPG Rise Climate Fund.

4. Neros Technologies, $250M, defense: Neros Technologies, a defense drone manufacturer and drone interceptor, raised a $250 million Series C led by American Strategic Technology Fund and Sequoia Capital. The Los Angeles-based company, founded in 2023, has raised $370 million to date, per Crunchbase data. The company announced that by 2028 it will be building 1 million drones per year. It has contracts with the U.S. military and half a dozen allied countries across Europe, Asia and the Middle East.

5. CodeRabbit, $143M, AI code review: CodeRabbit, which has developed an open-source AI code review technology used by 150,000 open-source projects and 17,000 customers, according to the company, is on a roll. London-based Atomico and Los Angeles’ Smash Capital led its Series C, and the Walnut Creek, California-based company plans to set up a London office based on customer interest across Europe. As coding becomes increasingly automated, code review is critical to determine which projects meet production standards.

6. Point2 Technology, $136M, datacenter networking: Point2 Technology, which builds interconnect solutions for AI data centers, raised an extension of its Series B led by Korea-based LB Investment, including a strategic investment from Arm Holdings and participation from existing investor Maverick Silicon. The San Jose, California-based company raised $136 million in the Series B.

7. Bridge to Life, $110M, organ preservation: Bridge to Life, a 21-year-old organ preservation company, raised a $100 million Series C funding led by Soleus Capital and Lauxera Capital Partners. The Illinois-based company is developing an organ viability assessment tool to expand the market and will use the proceeds to extend its VitaSmart technology to U.S. transplant centers.

8. (tied) Aureka Biotechnologies, $100M, protein therapeutics: Aureka Biotechnologies, an AI-native drug discovery platform, raised $100 million in Series B funding led by Granite Asia. Based in Shanghai and California, the 3-year-old company has raised $200 million to date. Proceeds will go toward training its model with its lab-in-the-loop feedback mechanism.

8. (tied) PointsKash, $100M, loyalty rewards: PointsKash, a loyalty reward mobile app, raised private equity funding led by Hawk Capital Investors. The Florida-based company provides in-store kiosks for cash, loans and rewards.

10. Epicrispr Biotechnologies, $90M, biotechnology: Epicrispr Biotechnologies, a developer of genetic medicines, raised a $90 million Series C led by Janus Henderson Investors and Octagon Capital Advisors. The San Francisco-based company develops treatments for neuromuscular disease through modulating genes.

Anthropic CEO Dario Amodei recently pushed back against the idea that he’s been painting an overly pessimistic picture of artificial intelligence and how it might shape the future.

Amodei’s comments came in response to investor Gavin Baker, who argued — both on the All-In podcast and on X — that Amodei’s warnings about the dangers of AI have helped to fuel a backlash in the United States, particularly against data centers.

Claiming that Amodei has “lost the argument” when it comes to AI regulation (Anthropic has advocated for some regulations, including a California bill that imposes transparency requirements on large AI companies), and given that “he is about to be the CEO of one of the most important companies in the world,” Baker wrote, “I respectfully think he should make an effort to be a more positive advocate for his own industry.”

Baker is far from the only one arguing that AI skepticism and even government crackdowns stem in part from simply taking the dire warnings of some AI executives seriously. But in his response, Amodei disagreed with the idea that his “messaging has been disproportionately negative.” Instead, he said that his writing has been “about equally balanced between risks and benefits,” and that he wrote his essay “Machines of Loving Grace” because he “didn’t feel the AI industry was painting an inspiring enough picture of how the technology could radically transform the world for the better.”

Nonetheless, Amodei acknowledged that “the public has a negative view of AI” and he agreed that “this is a big problem.” Where he disagreed was with the idea that this is “primarily caused” by Amodei “or any other AI leader warning about AI’s risks.”

“I think it is fundamentally a crisis of trust,” Amodei said. “I think that ordinary people don’t trust companies, governments, or the tech industry and always suspect that we are cooking up some new way to screw them over.”

Indeed, “trust” is a word that often comes up in debates about the AI industry, especially around OpenAI CEO (and Amodei’s rival) Sam Altman. In Amodei’s telling, however, this is a crisis that’s been decades in the making, with the AI backlash “just the latest iteration of it.”

“I think by far the most accurate criticism of AI companies including Anthropic is that we haven’t yet delivered on our big promises to benefit the world,” Amodei said. “That is totally on us, and I think it’s the criticism you should be making, instead of all this stuff about messaging and marketing.” (Naturally, he also said Anthropic is “doing our best to fix this.”)

As for regulation, Amodei argued that Baker was painting “a false choice” between distributing AI widely without regulation, or concentrating the technology in the hands of a few companies through regulation.

“I know that there’s a sort of Silicon Valley shorthand where regulation = regulatory capture = concentration of power, but I’ve always found this to be an overly simplified picture of the world,” Amodei said. “Many people outside this bubble think of regulation as something that constrains corporate power and benefits ordinary people.”

Amodei added that he doesn’t “necessarily agree with that perspective either,” but he said that’s “why Anthropic has always made its policy proposals very carefully.”

“We try very hard to make proposals that disadvantage (slow down) frontier AI companies while *advantaging* smaller competitors,” he said.

Some of the artificial intelligence boom’s biggest beneficiaries can cash in whether open models proliferate or not

A wave of cheap and powerful Chinese open-weight AI models is causing a scare among investors in the AI boom. For a decent chunk of the tech universe, however, there is less to worry about than meets the eye.

The rapid advance of open-weight models, which make public a set of numerical values governing their behavior, is certainly a concern for closed-model developers. Those companies don’t allow changes to how their models operate, giving them tighter control over how they respond to queries.

The leaders in that space, OpenAI and Anthropic, are both planning initial public offerings in the coming months. Any doubt about the durability of their competitive advantage could diminish their appeal to investors.

Google is another closed-model developer with something to lose. Gemini models have fallen a bit behind the cutting edge performance-wise in recent months; better and cheaper open-weight models threaten to pull customers away.

Worry about the impact sent Google parent Alphabet’s GOOGL -0.13%decrease; down pointing triangle stock 10% lower in the days after the debut of the open-weight Kimi K3 model from China’s Moonshot AI last month. Microsoft MSFT -0.30%decrease; down pointing triangle, Amazon.com and other AI-infrastructure heavyweights also fell.

Kimi K3 performed on par with, or better than, some leading closed-weight models, and access to it was priced far more cheaply. Other powerful and efficient models have emerged in China this summer, too, including ones from tech giant Alibaba and startups Z.ai and MiniMax 100 -12.69%decrease; down pointing triangle.

That has added to concern about U.S. leadership in AI and Big Tech’s insatiable spending on AI infrastructure. If Chinese companies can do leading-edge AI better and more efficiently with open-weight models, the reasoning goes, plans to spend untold sums on data centers and chips might need reworking.

The opposite is just as likely to be true if open-weight models take off, though. Should AI become far cheaper to deploy, the so-called Jevons paradox, named after a 19th-century economist, would likely take hold. That would mean people respond to lower costs simply by using the technology more, eating up just as much if not more computing power.

That is one reason chip companies have largely welcomed open-weight models. Andrew Feldman, the chief executive of the Nasdaq-listed AI chip company Cerebras Systems CBRS -5.21%decrease; down pointing triangle, said open-weight and closed-weight models pushing each other would benefit consumers, but wouldn’t hurt chip demand. “There is no reason for chip stocks to go down when open-source models come out,” he said.

Nvidia NVDA -0.06%decrease; down pointing triangle Chief Executive Jensen Huang has made himself an open-weight poster child in the past month, leading a consortium of companies championing the approach and drawing somewhat-dubious connections between open-weight models and the highly successful open-source software movement.

Beyond the chip industry, cloud-computing leaders like Amazon, Microsoft and even Google would also benefit handsomely if open-weight models expand demand for computing power. Open-weight models might be cheaper and more efficient, but while they might be free for anyone to pick up and use, they aren’t cost-free to run. That is a win for anyone in the computing-infrastructure game.

As Morningstar analyst Malik Khan said in a recent note, “if an enterprise were to consolidate its entire AI stack on open-weight models, it would still need cloud infrastructure to run those workloads, store data, manage security and access to resources, et cetera, all tailwinds to cloud infrastructure companies.”

A large number of corporations are already using open-weight models. They often tune them to excel at narrow tasks, like summarizing documents or answering customer questions on company-specific topics.

That can be a big money-saver compared with paying usage fees for smarter but more expensive closed-weight models. A McKinsey survey last year found nearly two-thirds of companies that had experience with AI were using open-weight models—mostly less than cutting-edge ones offered by Meta Platforms META -0.86%decrease; down pointing triangle, Google and France’s Mistral.

Yet there are reasons to doubt advanced open-weight models—particularly those from China—will make major inroads in the AI race. While there is plenty of demand, there is no obvious profitable business model for them. Which is one reason open-weight model developers haven’t attracted a lot of venture-capital interest.

Open models can make sense for big tech companies for whom they are loss leaders, costing them money but drawing users to other profitable parts of their businesses. It is less clear whether independent open-model developers can make enough revenue to pay for the gargantuan cost of training a cutting-edge model.

Cheaper pricing for Chinese models was “a business-model choice, not necessarily a hardware-cost read-through,” BofA Securities analyst Vivek Arya said in a note recently. Despite its efficiency gains, Moonshot’s Kimi K3 model still needs around the same amount of high-cost memory as an advanced OpenAI open model, he said.

The economics might not matter as much in China, given that all of its leading AI developers have state support and AI is seen there as a strategic priority. But a large-scale migration to China’s open-weight models, at least among large Western companies, seems unlikely even if they do remain cheaper.

DeepSeek, the Chinese AI developer that shook markets last January after releasing its own advanced model, quickly faded as a threat to the AI boom. This time could easily be similar, with the possibility of U.S. restrictions on the use of Chinese open models serving as a deterrent for prospective customers—even if the curbs never come.

The disruptive possibility of open-weight models appears likely to linger, especially for OpenAI and Anthropic. But for others, their impact looks far less severe.

They discovered the gold coins and bars while laying sewage pipes on the site of a former brewery in Dendermonde

Patients, doctors and nurses are turning to the technology for help identifying rare and undiagnosed diseases

Patients with rare diseases often spend five years or more searching for a diagnosis. Now artificial intelligence is helping some get answers faster.

Rachel Hinken long wondered why her son, Oliver, seemed to be missing key growth milestones, including speech and walking delays. He consistently tracked between the zero and first percentile for height. At age 10, he wasn’t more than 4 feet tall. Doctors said he was fine. He would catch up. Hinken wasn’t convinced.

Last year, she uploaded Oliver’s picture into Face2Gene, an AI-assisted app designed for healthcare professionals that uses facial analysis to flag rare genetic disorders. A strong match came up for Oliver: a type of Trichorhinophalangeal syndrome, or TRPS. The condition can cause bone or joint issues that lead to pain and problems with movement.

Genetic testing and clinical consultations confirmed the tool’s result, and revealed that Hinken had TRPS, too. “I screamed, I was like, ‘Holy crap! We finally have an answer!’” said Hinken, who lives on New York’s Long Island.

Patients and families are turning to AI to help pinpoint conditions. AI can be especially adept at flagging potential rare and hard-to-diagnose diseases, which may otherwise go undetected for years because doctors don’t often see them.

“I have been a little slack-jawed a couple of times by the difference” AI has made in helping to spot uncommon conditions, said Dr. Sarah Diekman, a doctor who specializes in a subset of nervous-system disorders like POTS. “I’ve seen patients who’ve had this for 50 years, and I was the first person to diagnose.” Now, she said, patients who identified POTS with the help of AI chatbots are coming to her within months of their first symptoms.

Fidji Simo, the former chief executive of Instacart, has been public about her battle with POTS and stepped down from her post as OpenAI’s No. 2 executive in July, citing her health. In an interview with The Wall Street Journal, she said she believes today’s AI tools would have helped her get a diagnosis, which took roughly nine months, much more quickly.

“I was a textbook case, and doctors missed it because they were just so focused on the fact that I had had a rough pregnancy,” said Simo, who first started having symptoms after giving birth to her daughter in 2015.

She has used ChatGPT to analyze the results of her whole-genome sequencing, gather information on experimental treatments and brainstorm potential tests to discuss with her doctor. Simo, who is still a part-time adviser at OpenAI, founded a new company focused on researching biological drivers of conditions like POTS and matching patients to drugs.

Data-center-driven diagnoses can be even more accurate than doctors’. In a study published last year in a JAMA journal, researchers took 90 complex rare-disease cases that had already been solved and tested AI chatbots’ abilities to suggest the right diagnosis. The two chatbots got the diagnoses right in 13% and 10% of cases, respectively, compared with 5.6% of cases that doctors had diagnosed through a review of patients’ medical records.

AI can be especially good at finding patterns, such as links between physical features in medical imaging and words in medical literature or case reports, doctors and researchers said. That can help spot rare diseases, especially in rural or nonspecialty clinics where doctors may be confronting them for the first time, said Dr. Matthew G. Hanna, a breast pathologist who chairs the College of American Pathologists’ AI committee.

AI may help provide suggestions for them about what types of tests to run, or what potential treatments could help, Hanna added.

Yet there are limits, doctors said. Clinical geneticist Dr. Xiao P. Peng, who helped confirm Hinken’s family’s diagnoses, said the tools still make mistakes and are better at generating possible leads, translating complex medical jargon and compiling research than making diagnoses.

“It searches much faster than we can across many more spaces,” said Peng, who is director of advanced diagnostics and therapeutics at the New York Center for Rare Diseases. “But for data synthesis, there has to be a manual human role.”

Michael Ames, a nurse practitioner at Mayo Clinic in Rochester, Minn., used an internally developed, now FDA-cleared AI tool that helped lead him to further testing that confirmed an unexpected diagnosis for one of his patients.

The patient, Mike Busch, a 77-year-old insurance agent, went to Mayo with symptoms of shortness of breath and feeling like a weight was on his chest. He suspected pneumonia. His care team suspected heart failure. He might have been treated for something else, said Ames, were it not for an AI interpretation of Busch’s electrocardiogram that suggested a 98% probability of cardiac amyloidosis, a rare, serious heart disease.

“I was fairly shocked,” said Ames, who had never diagnosed the condition before. He was also skeptical about the AI’s accuracy. Yet further imaging confirmed the diagnosis.

Busch, who said he had never taken a regular medication in his life before the diagnosis, now takes seven to 10 a day. He is also working on getting more exercise and eating less salt.

“If it wasn’t for AI, I might have been treated for something else,” said Busch. “Maybe I wouldn’t be here, who knows?”

The broader problem with AI in rare-disease diagnosis is that existing tools are only as good as the data they are trained on, researchers, doctors and nurses said. There are fewer data points, including case records, tissue samples and images, for rare diseases than more common conditions. Those that do exist aren’t often well digitized.

Hanna helped develop a mobile shipping container called ScanVan that physically travels to the Department of Veterans Affairs hospitals and academic centers to pull data from archives and make it available. The goal is to use them to build a better, more accurate AI model for rare-disease diagnosis.

Anthropic said in July that it plans to start funding biotech research on drug development and diagnosis for rare diseases—a process the AI company says it believes its flagship Claude product can speed up.

Pharmaceutical companies are getting in on the action, too. Consulting firm ZS has built tools for drugmakers designed to screen patients who may have rare diseases, said Bill Coyle, ZS’s region managing principal for Europe. The firm helped build, for example, a tool that screens for a rare neuromuscular disorder called myasthenia gravis. Some 20 of 140 people who used the tool later reported being diagnosed with the condition, Coyle said.

“We’re seeing more demand for this kind of topic because it’s critically important if you think about launching a new product,” he added.

AI could also help patients organize their medical records and point them to relevant specialists, said Danielle Carnival, CEO of the nonprofit Undiagnosed Diseases Network Foundation. Yet the technology could also create more confusion for patients if it produces more information without a clear, accurate path forward, she cautioned.

“We put too much responsibility on the families to quickly become experts in what they’re facing,” Carnival said.

Just 13% of U.S. large-cap funds have outperformed indexes over the past decade, new data show

Investment pros say AI disruption has created a stock picker’s market. They are still struggling to pick the right ones.

Just 27% of actively managed U.S. large-cap equity funds beat their benchmark passive-fund alternatives in the 12 months ended June 30, according to Morningstar data that compares returns after fees. That is actually an improvement over stock-picking funds’ long-term track record. In the decade through June, just 13% of active large-cap funds beat their benchmarks.

The fees on actively managed mutual funds have been a cash cow for investment firms for decades, but years of outflows from investors moving to more tax-efficient, and often passive, exchange-traded funds are weighing heavily on the industry.

“If you look at active equity mutual funds, they’ve had outflows every year consistently since 2015,” said Matthew Bartolini, head of SPDR Americas research at State Street Global Advisors. “That’s a losing trend only compared to the New York Jets.”

In recent Wall Street marketing and outlook materials, stock picking is still all the rage. Active managers argue that higher interest rates have ended the cheap-money tide that lifted all boats, just as AI is set to mint huge winners and losers, making security selection crucial.

“Conditions have shifted to favor active investing,” T. Rowe Price declared. AI means “active stock selection will be increasingly important,” says Janus Henderson. “Finally, active managers can participate and join the passive moneymakers,” Jefferies CEO Rich Handler said in a recent letter to clients.

Those predictions have been correct, to an extent. Stock-market dispersion—or the divergence in performance between individual stocks in the same index—has soared to the highest level in decades this year.

That should theoretically make it an ideal environment to beat low-fee index investing by picking winners. But the S&P 500 and Nasdaq-100 are weighted by market value, and outsize gains from a handful of superstar companies continue to drive returns. The S&P 500’s 10 largest companies represent more than 40% of its value, according to Dow Jones Market Data—the highest concentration since the 1960s.

Many active managers are uncomfortable making bets that concentrated, especially since nearly all of the value is tied to the technology sector.

“That amount of concentration is broadly viewed as too deep for most portfolios,” said Holly Framsted, head of product group at Capital Group, one of the largest active-fund managers. “It’s really important to recognize that if you get the direction of travel on that theme wrong, the risk is outsized.”

Investors should consider both diversification and the total portfolio risk in their portfolios when comparing active and passive results, Framsted said.

U.S. index-tracking funds pulled even with active funds in total assets for the first time in 2020, according to the Investment Company Institute, and now hold almost twice as much money. That is a far cry from the pre-financial-crisis landscape: In 2007, assets in active equity funds outnumbered those in passive strategies by more than three to one.

Still, the proliferation of passive funds that track nearly anything, from total-market indexes to small subindustries, means plenty of individuals and institutions use such funds as tools to make their own active decisions, like going overweight on a specific theme.

Low-cost passive ETFs are on pace to hit $1 trillion in net inflows for the first time this year, underscoring the dominant role they now play in portfolios.

Bonds have been more of a bright spot for fund managers. Active intermediate core bond funds, the largest category, had a 66% success rate beating benchmarks over the past year, and a majority have done so for three years running, according to Morningstar. Active fixed-income ETFs have been growing faster than passive funds.

Investors considering active management would be wise to look beyond large-cap stocks, State Street’s Bartolini added.

“You can go get stock-market beta with ETFs in a very fee-efficient and tax-efficient manner, and then use your active budget elsewhere where there might be more opportunities, like fixed income,” he said.

Big Manufacturers Find New Demand in Equipping AI Data Centers

Companies such as Caterpillar and Cummins are pivoting to feed a booming market for once-prosaic power equipment

U.S. manufacturing is booming as industrial companies pivot to meet demand from AI data centers.

Caterpillar and Cummins are investing hundreds of millions of dollars to expand their electricity generator production.

Ford plans to redirect EV batteries to data centers and other industrial users, while Eaton has been acquiring businesses to expand its data-center operations.

U.S. manufacturing is booming thanks to AI data centers, and industrial companies such as Caterpillar CAT 0.23%increase; up pointing triangle, Eaton ETN -0.40%decrease; down pointing triangle and Ford Motor F 3.46%increase; up pointing triangle are pivoting their business to seize the moment.

Manufacturing last month rose to its highest level since 2022, when the recovery from the Covid-19 pandemic fueled a factory-production frenzy. Today, data centers and a handful of AI-related industries are driving the manufacturing sector, and companies in that sector are investing hundreds of millions of dollars to expand their lineups to capitalize on the opportunity.

Power-hungry data-center developers have turned Caterpillar’s once-prosaic business of electricity generators into the equipment maker’s leading source of profit. The company is investing $725 million to expand generator production at an Indiana plant.

Caterpillar converted another plant in Kansas to produce turbine engines popular with data centers. The company is also resuming production of 10-megawatt generators that were last made in 2022.

“If we can get more units out, they’re asking us to give them more,” Caterpillar Chief Executive Joe Creed said of data centers during the company’s second-quarter earnings call with analysts.

Engine maker Cummins CMI -0.06%decrease; down pointing triangle is also leaning in to its existing generator business. The company is investing $450 million to increase generator production, following a $200 million investment in expanding production completed last year. Cummins expects its data-center-related sales to rise by 80% to $9 billion in 2030 from 2026.

Cummins’s generators are now mostly used for standby power in data centers. Data centers’ voracious power consumption is pushing developers toward on-site power generation and storage. Cummins this spring said it plans to offer new larger generators starting in 2028 that can be grouped at data centers as a primary power source. The generator’s 130-liter engine—roughly 65 times the average size of a U.S. car engine—will be powered by natural gas and produce four megawatts of electricity.

At the same time, economists warn that aggressive spending on data-center construction and equipment increases the risk that the build-out and the costs will overshoot the demand for data-center capacity.

For now, many companies are moving forward carefully. Jenny Bush, president of Cummins’s power-systems business, said the Indiana-based company is expanding production in smaller increments and limiting plant expenses.

“Everybody is naturally worried about when the build-out gets completed,” she said. “We’ve been trying to utilize as much of what we already have before we build anything new.”

Ford Motor is looking to data centers to repurpose excess production capacity for electric vehicle batteries. The automaker’s new Ford Energy subsidiary expects to spend $2 billion redirecting batteries to electricity storage at data centers and other large industrial users of power. Demand for electric vehicles is in a slump, leaving automakers with the need for fewer batteries than originally anticipated. Ford’s interest in energy storage comes as its auto business is smaller than it was a decade ago.

Manufacturing in support of the data-center business is helping to offset market conditions that are holding back other industrial sectors. Inflation, high interest rates and rising material costs are slowing down construction, auto sales and demand for consumer durable goods.

Rising demand for steel and aluminum and a 50% tariff on imported metal have handed domestic producers unprecedented pricing power this past year. Tariffs have pushed U.S. prices for steel and aluminum to among the highest in the world.

Higher prices for Caterpillar’s equipment and generators aren’t holding back sales. The company said price increases contributed nearly $600 million to its second-quarter operating profit, which was 50% higher than a year earlier. The Texas-based company earlier this month reported a record $20.5 billion in quarterly revenue.

Caterpillar shares are up about 50% since the start of the year. Its backlog of generator orders from data centers stretches as far out as 2030, Caterpillar said.

“No one is slowing down at the moment,” Creed said.

Eaton, a century-old manufacturer of circuit breakers, switchgear and other electrical equipment, has spent $13 billion on acquisitions since 2025, including businesses that have expanded its data-center operations. Data centers and distributed IT accounted for 21% of the company’s sales last year, up from 14% at the end of 2023. Eaton’s second-quarter sales overall rose 21% from a year earlier.

“What we have ahead of us in terms of demand is still incredible,” Chief Executive Paulo Ruiz said late last month.

The Institute for Supply Management’s purchasing index, a broad gauge of manufacturing activity, has been above the threshold for manufacturing-sector expansion for seven straight months.

But high costs are starting to take a toll on some industries. The ISM reported that 20% of the manufacturing GDP last month had an index reading that signaled contraction, up from 5% in June.

The group said the chemical industry accounted for most of the contraction, attributing the weakness to the rising price of oil, which many companies use as their primary feedstock for chemicals.

Cost pressures and oversupply risks exist for data centers, said Willy Shih, a management practice professor at Harvard University. Evolving technology could make future data centers smaller, cheaper and less power-hungry. He said the current build-out reminds him of the dot-com era of the late 1990s, when huge investments were made in building fiber optic networks that remained unused for 20 years.

“The inevitable correction to this is that buyers will find a cheaper way and companies that put on too much capacity end up with a boom-and-bust cycle,” Shih said.