TheChek News Flow

UK struggling to reach state aid agreement with Chinese-owned company 18 months after plea for £500mn support package

The UK government is struggling to seal a deal with British Steel to secure the future of the country’s second-largest steelworks 18 months after the Chinese-owned company asked ministers for a £500mn financial aid package.

Jingye, which bought Britain’s second biggest steelmaker out of insolvency in 2020, is yet to submit a detailed, fully costed proposal that can be put before the Treasury for approval, according to people familiar with the matter.

The potential aid package would keep steelmaking alive at British Steel’s flagship Scunthorpe site by funding a switch to greener forms of steelmaking that would protect some jobs.

“There are definitely some jitters right now, this was meant to be agreed around the New Year,” said one government figure. “The pace of the talks has visibly slowed, it’s not clear how much progress is happening.”

Industry and government figures said talks were continuing and that an agreement could still be struck by the summer. Two people close to the situation said Jingye would soon submit a full proposal.

A deal for the Scunthorpe site would mirror a similar agreement with India’s Tata Steel plant at Port Talbot in south Wales.

Tata Steel is receiving £500mn of state support to help its transition to producing green steel through electric arc furnaces, which are less carbon-intensive and require fewer workers.

Up to 2,800 Tata Steel jobs are set to be lost. Tata’s national-level consultation with unions is expected to end next week, when the company could confirm it will close its two remaining blast furnaces as soon as September. Unions have been balloting members over industrial action about the proposed cuts, with Unite voting in favour last week. 

Likewise British Steel said last November that it would also close its two blast furnaces and instead build two electric arc furnaces, including one at its Teesside site, meaning up to 2,000 job cuts.

The company said the restructuring would require £1.25bn of investment and would be contingent on securing some state support.

Britain’s two remaining blast furnace operators have long warned they need help to cover the costly shift towards less energy intensive forms of steelmaking. At the same time, they have suffered from high energy costs compared with continental rivals and challenging market conditions.

British Steel has previously said it was losing £1mn a day. Its auditor resigned in January after warning that there was “material uncertainty” over the company’s ability to continue operating without another injection of funding from Jingye.

In October 2022 the Financial Times reported that Jingye wanted £500mn of government support. Ministers have offered a lower figure of £300mn.

Ministers are expected to demand similar conditions from British Steel that they won from Tata, including long-term promises on research and jobs. 

British Steel said the “process is ongoing to finalise detailed plans as per scheduled timescales”.

The company added it remained committed to the UK and said owner Jingye “continues to stand resolutely by our side after ensuring we overcame the significant financial challenges posed by Brexit, the pandemic, and the global economic crisis”.

While talks with the UK government were continuing, the company warned that given challenging market conditions, “it is imperative swift and decisive action is taken to ensure a sustainable future for British Steel”.

The Department for Business and Trade said: “We have offered British Steel a generous support package including more than £300mn of investment for them to cut emissions, help safeguard jobs and create a positive future for steel production. Discussions with the company continue.”

Rules are tightening and prices rising but plenty of schemes remain open for business

Where is a millionaire to go these days? Spain is nixing its property-linked “golden visas”, joining Portugal and Ireland. The UK is scrapping its 100-plus-year-old “non-dom” regime enabling wealthy foreigners to skip tax on overseas income. The opposition Labour government vows it would crack down on non-doms even harder.

Sovereigns used to see selling residency or citizenship rights as a way to attract foreign investment. Island nations were early proponents, with St Kitts and Nevis offering citizenship via investment in 1984. In Europe, countries including Greece, Spain and Portugal joined the club about three decades later, in some cases to prop up flagging property markets.

But unintended consequences have come to the fore in recent years. As Spain and Portugal discovered, influxes of foreign money distort property markets, edging out domestic buyers and tenants.

Authorities also balked at the scope for more nefarious activities. The EU, flagging concerns including organised crime, money laundering and tax evasion, is trying to call time. Bowing to pressure, several Caribbean countries — some of which have passports granting visa-free access to European countries on short trips — plan to double the minimum investment requirement this year to $200,000. They have also agreed to improve transparency and regulation and beef up security checks.

Yet demand for second bolt-holes is rising along with wealth. No wonder. This, after all, is the ultimate bulwark against economic and political uncertainties, much sought after at a time of multiple elections, geopolitical tensions and regional wars. Wealthy Chinese investors, long fans of Caribbean passports, are the biggest group of those taking up the Portuguese visa. Russian investors were also buyers, until blocked by European countries following the war in Ukraine.

Henley & Partners, which advises wealthy individuals on the subject, forecasts a 15 per cent increase in the number of applicants to golden visa schemes to 128,000 this year. Rules are tightening and prices rising but plenty of schemes remain open for business. Italy, Greece and Malta are all attracting applicants. US millionaires, along with their wealthier centi-millionaire and billionaire brethren, are leading the charge.

That may be a feature of scale: roughly a third of global investable wealth — defined as cash, listed company holdings and debt-free property — resides in the US.

For sure, tighter scrutiny and rising investments will be an ongoing feature of schemes. But mounting wealth piles and the desire for bolt-holes means golden visas are likely to be around for quite a while yet.

A glut of apartments available in the French capital this summer is pushing down prices

Parisians aiming to cash in on the Olympic Games by renting out their apartments are struggling to lock in bookings as a glut of listings comes on the market, pushing down prices.

Only about one-third of available Airbnb rentals for the Paris area have so far been booked during the Olympics, according to data analytics group AirDNA, while 3,000 to 3,500 new listings are coming online each month.

Many Parisians intend to flee the city during the games, which start on July 26, as an expected 15mn people visit the capital, according to tourism officials. However, their plans to rent their homes to tourists at high prices during the sacrosanct French August holiday period are being dashed as supply outstrips demand.

Stefania, a banker living in Paris’s trendy 10th arrondissement, has been renting out the studio next to her apartment on Airbnb for the past year and half. “Normally it books very fast, but I opened it a month ago for July and August bookings and so far there are none,” she said. 

She usually charges €150 a night, but for the two-and-half weeks of the Olympics she had put it up to €250. The lack of bookings “seems unusual for what I thought would be a period of heavy demand, but that €100 is not going to make or break my day”, Stefania said.

“I figure it will rent eventually but we might have to lower the price.”

She is not the only one having to reconsider pricing. While the average price being asked by would-be Airbnb hosts in Paris is €594, the average nightly rate for booked accommodations for the Olympic period is €323 according to AirDNA.

Airbnb, which is also an official partner for the Olympics, remains optimistic about the opportunity it offers hosts.

Tens of thousands of new hosts have listed accommodations in the cities hosting Olympic events since the start of 2023, according to the company. A Deloitte study commissioned by the company estimates that the average host in the Paris region will generate €2,000 in additional income during the games. 

“Paris 2024 is set to be the biggest hosting event in Airbnb history, with more guests staying in local homes on our platform than at any event, ever before,” the company said. “Thousands of people in host cities have opened their homes for the first time, and more than half of listings that have received a booking did so in less than seven days.”

But securing those bookings remains challenging, as blocks of hotel rooms previously set aside for Olympic delegations also come on the market.

Hotels have started to drop prices as well, as they compete for occupancy, according to tourism specialist Olivier Petit at consultancy In Extenso. While London built around 7,000 new hotel rooms for its Olympics in 2012, Paris has only added around 2,000 rooms, according to Petit.

“In the short term this could be a bit challenging for hotels . . . but in the medium term it’s much better to have a flexible stock of Airbnb oversupply as opposed to permanent expansion of hotel room stock that constantly needs to be filled,” he said. 

Many would-be Olympic rental hosts are likely to be disappointed. There has been “an explosion of the supply” across holiday rental platforms including Airbnb, Abritel and Booking.com, according to Lycaon Immo co-founder Stéphane Daumillare, who says there are currently about 15,000 available bookings across these platforms. 

“Those who had a lot of budget and were organised have already booked during the period when prices were very high in November and December, making prices go up,” he said. 

Now as the offer explodes, “we’ve seen prices coming down for the past three months”, said Daumillare. He estimates that only about one in four of the listings available will be booked for at least three nights during the Olympics. 

“The market is becoming more rational. The train has left the station for extremely high priced bookings,” Daumillare said. However, “there will still be opportunities for later bookings because in addition to normal tourists, there will be people who come for the Olympics and then stay for tourism.”

Air Products and Chemicals has had a tough start to the year, but its stock is now too cheap to ignore.

“Tough” might not be doing Air Products’ decline justice. Shares have dropped 15% in 2024, to a recent $233.02, with a big chunk of that loss coming in February after the chemical company reported first-quarter earnings that missed expectations by a wide margin. Management also cut its full-year earnings guidance to $12.65 at the midpoint of its range, while noting that weakness in China was weighing on revenue.

That puts the second-quarter report—out April 30—in the spotlight. Analysts are forecasting sales of $3.05 billion, down from $3.2 billion during the same quarter in 2023, with earnings dipping to $2.70 a share, down from $2.74, as falling production costs are outweighed by higher interest expenses. Numbers like that would put the company well short of its full-year guidance, but the market seems OK with that—analysts have projected rising sales and earnings during the third and fourth quarters of this year.

That narrative seems reasonable. Analysts expect sales for the full year to grow just 1.8%, with Asia industrial gases’ revenue projected at 3.2% growth, not a big reach given that China’s economy is still growing and its first-quarter gross domestic product grew faster than expected. Plus, the company has traditional hydrogen projects in other regions, such as Saudi Arabia and Uzbekistan, that should add revenue. These projects have just begun to generate sales and are expected to continue to ramp up for the rest of this year.

“We expect incremental volume growth as some of these projects continue to ramp, which should help to offset the ongoing headwinds in Asia,” writes Wells Fargo analyst Michael Sison.

Investors will also be watching to see what the company says about the demand for clean hydrogen. In its 2023 10-K, released in November, Air Products said that it will invest more than $5 billion this year in new projects devoted to the fuel. That’s a lot to spend on a chemical that has essentially no demand at this point. Still, McKinsey estimates that sales of tens of millions of tons by 2030 isn’t out of reach. Any sign that the company is locking in contracts and strong pricing with clean-hydrogen customers could lift the stock, according to Vertical Research Partners analyst Kevin McCarthy, who calls clean hydrogen a key focal point.

All this isn’t a perfect setup for Air Products shares, but valuation is working in its favor. At 17.8 times 12-month forward earnings, the stock is rarely this cheap. Its current price/earnings ratio is well below its five-year average of 24.5 and near its lowest level since 2016. That’s a lot of bad news reflected in the stock.

Just a little bit of good news on April 30 should send shares higher.

Samsung Electronics is going big in Texas—real big. The South Korean tech giant got a U.S. Commerce Department Chips Act grant of up to $6.4 billion this week, to fuel investment north of $40 billion just outside Austin—if all goes according to plan.

Samsung’s U.S. ambitions echo those of rival Taiwan Semiconductor Manufacturing in neighboring Arizona. But investors are treating the two companies much differently.

Samsung has been the laggard as artificial-intelligence fever sweeps the global semiconductor industry. Its shares have gained 20% over the past year, compared with a 60% jump for TSMC, which dominates the manufacture of so-called logic chips. Samsung’s chief competitors in memory chips, fellow Korean SK Hynix and Idaho-based Micron Technology, have both nearly doubled.

There’s good reason for the underperformance. Samsung still runs a distant second to TSMC in logic, despite pouring some $50 billion into its foundries over the past few years. It ceded the lead in cutting-edge high-bandwidth memory to SK Hynix, which is supplying the most advanced designs of industry darling Nvidia

“Samsung is firmly in last place in HBM technology and well behind TSMC in logic,” says Dylan Patel, chief analyst at SemiAnalysis.

Some investors like Samsung stock anyway.

Cutting edge isn’t everything for a company pushing $200 billion in annual revenue. Samsung’s bread-and-butter DRAM and NAND memory chips should rebound this year after a glut that cratered prices up to 70% in 2022-23, says James Lim, a partner at Dalton Investments.

“Memory will allow Samsung to continue outgrowing the broader semiconductor market,” echoes Andrew Keiller, an emerging markets investment specialist at Baillie Gifford. “We see 8% to 10% revenue growth at slightly improving profitability.”

Governance at Samsung also looks better than average for a Korean chaebol, though that’s admittedly a low bar. Executive Chairman Lee Jae-yong (anglicized as Jay Y. Lee), who still calls the shots for the founding family, was acquitted of his latest criminal charges—manipulating stock prices in a 2015 subsidiary merger—in February.

Lee and his sisters are in their 50s, young enough not to be nudging Samsung’s stock price lower to minimize Korea’s 60%-ish inheritance tax, a common chaebol strategy, Lim says. On the contrary, they have pledged company shares in a settlement to pay their own taxes. They benefit as the stock rises.

Leads can change hands in the ever-evolving semiconductor field. Samsung is looking to leapfrog SK Hynix in high-bandwidth memory with so-called HBM 3E technology, due to roll out this year and next, and even challenge TSMC with “gate all-around,” or GAA logic chips. GAA chips at four nanometer thickness could be rolling off Texas assembly lines as early as 2026, Lim says.

Samsung’s greatest strength, though, may be that semiconductors are too essential to be a winner-takes-all industry, and too expensive and complicated to admit many competitors. The company can keep churning out profits and growth as a solid No. 2 across various product lines. “The whole world really needs Samsung to succeed right now,” says Igor Tishin, a tech analyst at Harding Loevner. “There’s too much reliance on TSMC and Hynix.”

Jensen Huang, the Nvidia CEO and tech demigod of the moment, seemed to underline this point himself last month, telling Nikkei Asia, “I value our partnership with SK Hynix and Samsung very incredibly.”

Samsung shares have climbed 11% since then. The game isn’t over for the Korean powerhouse yet.

Its merger with Japan’s Nippon is in doubt, but shares still look

The takeover of United States Steel by Japan’s Nippon Steel is in limbo following opposition from both Republicans and Democrats. Its stock is still a buy.

U.S. Steel’s history stretches back to the 19th century and includes names such as Andrew Carnegie and J.P. Morgan. Its blast furnaces—300-foot-high tubes with iron ore, coal, and limestone packed at the top and molten hot pig iron flowing out the bottom—once symbolized America’s industrial might.

No longer. U.S. Steel is now just the world’s 27th-largest steel company based on steel output, after falling from 12th largest a decade ago. It’s the third-largest steel maker in the U.S., trailing Nucor and Cleveland-Cliffs. U.S. Steel accounted for more than 10% of global steel production in 1950. In 2023, it accounted for less than 1%.

Many factors contributed to the decline. The company diversified into chemicals and oil in the 1980s, while Nucor’s pioneering use of steel scrap melted with electricity eroded U.S. Steel’s domestic market share. Steel also tends to get produced where it is needed. China’s rapid industrialization over the past 20 years has made it today’s global steel-making behemoth.

It isn’t easy to be a small, relatively high-cost player in a mature, commodity-oriented industry. Fewer shipments and lower profits become a downward spiral as diminishing cash flow leaves less money for capital spending. That’s another reason U.S. Steel keeps shrinking.

So it shouldn’t come as a surprise that U.S. Steel decided to explore its strategic options in August after rejecting a $35-a-share bid from Cleveland-Cliffs. The bidding, which likely included Nucor and steel service center Esmark, was eventually won by Nippon Steel, which offered to pay $55 a share in cash. U.S. Steel stock traded north of $50 in mid-December, up from $23 before the bidding started.

The deal makes sense. Nippon, the world’s fourth-largest steel maker, would bring an injection of capital and technology from a large global player, arresting U.S. Steel’s slow decline and positioning itself as Nippon’s production hub in North America. U.S. politicians, however, didn’t see it that way. Democrats and Republicans alike decried the deal, insisting that U.S. Steel must remain in U.S. hands. Comments from President Joe Biden panning the takeover on March 14 sent U.S. Steel shares down 18% over two days to $38.26.

The political fear is mainly about jobs. U.S. Steel employs some 22,000 workers globally, with 14,000 people in North America. About 11,000 of its U.S. workers are represented by unions. Today, U.S. Steel has more pension beneficiaries than employees—about 60,000. Overall, American steel mills employ fewer than 100,000 workers, according to the Bureau of Labor Statistics. There are about 165 million jobs in the U.S.

The U.S. steel industry is tiny. The market capitalization of the four largest U.S. steel producers is about $90 billion. Nucor, the largest American steel company, accounts for about half that amount. Production for the four amounted to about 68 million metric tons in 2023, more than 75% of the U.S. total but only about 3% of global output. The U.S. doesn’t make much steel anymore. Letting U.S. Steel struggle as a stand-alone company won’t change any of that.

On Wednesday, Biden called on the U.S. Trade Representative to raise tariffs on imported Chinese steel from about 8% to 25%. China is the world’s dominant producer, making about one billion of the 1.9 billion metric tons produced in 2023; it exports about 600,000 tons to the U.S. Imports accounted for some 20% of total U.S. demand last year, and a reduction in the supply or an increase in the cost of imported steel could cause prices in the U.S. to rise. Tariffs aren’t a primary reason to own U.S. steel stocks, but they don’t hurt.

U.S. Steel is probably worth more than the market is giving it credit for. At a recent $40.30 a share, its stock trades for about 12.4 times estimated 2025 earnings per share of $3.25, a small discount to the 13.5 times average multiple for Cleveland-Cliffs, Nucor, and Steel Dynamics. That’s up from 11.8 times forward-year earnings in August, before details of the Cliffs bid emerged. If U.S. Steel were to trade at 11.8 times after a Nippon deal break, its stock would fall to about $38, down 5%.

But a lot has changed since then. Steel prices are moving higher. That doesn’t hurt, and is one reason the sector is trading at a better multiple. Hot rolled coil, a benchmark steel product, currently costs about $850 a ton, up roughly $50 from last summer. Higher steel prices have shown up in Wall Street estimates. Analysts expect U.S. Steel to generate $1.8 billion in earnings before interest, taxes, depreciation, and amortization, or Ebitda, in 2024, up from $1.6 billion in August. What’s more, if U.S. Steel traded at the pre-Nippon discount to its three steel peers, shares wouldn’t budge.

And while a Nippon deal probably won’t happen, U.S. Steel could still be in play. Cleveland-Cliffs would likely still be interested in creating a larger, more-sustainable North American steel giant. It was willing to pay $54 in cash and stock for U.S. Steel—its final bid before Nippon offered more.

“We don’t see any scenario where U.S. Steel would take an alternative offer below our new $46 price target,” wrote Wolfe Research analyst Timna Tanners in a recent report.

And perhaps a whole lot more.

Amazon.com, Google, Meta Platforms, AMD, and Intel have big ambitions in AI chips.

Artificial intelligence has delivered seemingly daily wonders for the past 18 months. For investors, the biggest surprise has been the rise of Nvidia, which has come from humble roots to thoroughly dominate the market for AI-related chips.

Once known mostly for building PC add-on graphics cards for gamers, Nvidia has transformed its graphics processing units, or GPUs, into the beating heart of the AI revolution, powering the creation of large language models and running the inference software that leverages them in data centers around the world. Nvidia has been nearly alone on the field, with more than 90% market share.

But fresh competition is coming—from companies big and small—and the battle will be fierce. The stakes couldn’t be bigger: Lisa Su, the CEO of Advanced Micro Devices, has sized the AI chip market at $400 billion by 2027. Intel CEO Pat Gelsinger has projected a $1 trillion opportunity by 2030. That’s almost twice the size of the entire chip industry in 2023.

Nvidia’s Jensen Huang has built a company that is universally respected and admired, but chip buyers aren’t keen on relying on a single source. Hardware companies such as Dell Technologies, Hewlett Packard Enterprise, Lenovo, and Super Micro Computer can’t get enough Nvidia chips to meet customer demand—and they’d like alternatives. Cloud providers like Amazon.com and Alphabet’s Google want more options so badly that they are designing their own chips. And companies that rely on AI-based systems want more computing resources at more manageable costs than they can get now.

Nvidia’s success is now an opportunity for everyone else.

It’s hard to find a product of any variety that has had more impact on the financial markets so quickly than the Nvidia H100 GPU, which launched in March 2022.

Nvidia’s share price has more than tripled since the H100’s debut, boosting the company’s market value to $2.1 trillion. Among U.S.-listed companies, only Microsoft and Apple have higher market caps. And no other chip company is anywhere close.

This is no GameStop or Trump Media & Technology. In fact, Nvidia is the anti-meme stock: The company’s revenue growth has actually outpaced the stock gains. For its fiscal fourth quarter ended on Jan. 28, Nvidia posted revenue of $22.1 billion, up 265% from a year earlier. The company’s data center revenue was up 409%.

A few weeks ago, Nvidia launched its latest marvel, the Blackwell B200 GPU, which CEO Huang says dramatically outperforms the H100. With Blackwell, Nvidia raises the bar for its rivals. Nvidia for the foreseeable future will sell as many Blackwells as it can make—or, to be more precise, that partner Taiwan Semiconductor Manufacturing can make for it.

Huang has said Blackwell GPUs will cost $30,000 to $40,000 apiece. The current H100s sell in the same range. But the chip prices aren’t the whole story. AI customers want to run workloads in the shortest time, at the lowest cost, with the highest accuracy and reliability, drawing as little power as possible. There are a number of companies that think they can do that as well—or better—than Nvidia.

Nvidia’s rivals fall into three groups: big chip makers, cloud computing vendors, and venture-backed start-ups. With a $1 trillion market at stake, this won’t be winner-take-all. It isn’t game over. It’s game on.

Nvidia’s most obvious challengers are Advanced Micro Devices and Intel.

AMD shares have rallied 71% over the past 12 months, aided by the market’s perception that its new MI300 GPUs will chip away at Nvidia’s stranglehold on the market. That hope is inspired by AMD’s success at stealing market share from Intel in PC and servers.

“AMD is really the only other company on the field,” contends Andrew Dieckmann, general manager of AMD’s data center GPU business. “We’re the only other solution being adopted at scale within the industry.” He says that AMD chips outperform Nvidia’s H100 for many inference workloads, while offering parity for model training. But AMD’s other asset is that it isn’t Nvidia.

“For the very large users, they are not going to bet their entire franchise on one supplier,” Dieckmann says. “There is an extreme desire for market alternatives.”

AMD CEO Lisa Su said on the company’s most recent earnings call that she now expects 2024 GPU revenue of $3.5 billion, up from a forecast of $2 billion a quarter earlier.

Intel is coming from behind—and the stock has struggled for years—but the company opened eyes this month with the launch of Gaudi 3, its third-generation AI accelerator chips for training and inference. Intel contends that Gaudi 3 is faster than Nvidia’s H100 for both AI tasks, while using less power—and that Gaudi 3 will be competitive with Blackwell.

Intel, which is now spending billions to build out chip fabs in Arizona and Ohio, should be advantaged over the long run by having its own source of supply in a market with a severe supply shortage. But not yet: Gaudi 3 will be produced by TSMC, just like AI chips from Nvidia and AMD.

Jeni Barovian, vice president of Intel’s Network & Edge group, credits Nvidia for “establishing a foundation for this revolution,” but says Intel doesn’t intend to be left behind. Customers want alternatives, she says. “They don’t feel like they’re getting a choice today.”

Then there’s Qualcomm, another long-term leader in the semiconductor arena. The mobile-phone chip company has taken technologies originally designed for smartphones and applied them to the cloud in an AI inference chip it calls the Cloud AI 100. Qualcomm ultimately seems more interested in the opportunity to serve the edge of the networks, on laptops and phones.

Qualcomm senior vice president Ziad Asghar thinks that, over time, more inference workloads will be handled on “edge devices.” The theory is that it’s cheaper, and more data-safe, to stay out of the cloud. “The center of gravity in inference is shifting from the cloud to the edge,” Asghar says.

Less visible but no less serious about competing in AI chips are the internal teams at four cloud-computing giants—Amazon, Alphabet, Meta Platforms

META

-4.13%

, and Microsoft. All four are designing proprietary chips for both their own internal needs and to serve cloud customers. The competitive story here is less direct—none of the cloud leaders sell chips to third parties.

Nonetheless, they’re still a threat to Nvidia. Meta, Microsoft, Amazon, and Google together are spending a fortune on AI infrastructure. Those four are expected to have combined fiscal-2024 capital spending of $178 billion, up more than 26% from a year earlier. Microsoft alone will see capex increase 53% this year, according to estimates tracked by FactSet. Spending will jump 31% at Alphabet, and 26% at Meta. All four are building chips in part to gain better control of their spending—and they all say they can get there by controlling more of the “stack,” including software, hardware, and chip design—not unlike Apple’s approach to hardware design.

Nvidia declined to comment for this article, noting that the company is in its pre-earnings quiet period. But Huang did address the question of competition in the recent past.

“We have more competition than anyone on the planet,” Huang said at a March event at the Stanford Institute for Economic Policy. “Not only do we have competition from our competitors, we have competition from our customers.” His view is that the high levels of integration and efficiency built into Nvidia’s systems make it tough for rivals to keep up. Huang has said that the company’s total cost of operation is so good that “even when the competitors’ chips are free, it’s not cheap enough.”

That hasn’t stopped Big Tech from trying, and they’re being joined by a host of venture-backed start-ups. Some of those start-ups are hoping to sell chips and systems to server and cloud providers, but most of them are trying to disrupt the market by offering cloud-based services directly to customers.

The most intriguing of the group might be Cerebras Systems, which reportedly is planning a 2024 IPO.

In March, Cerebras unveiled its largest chip to date, the Wafer Scale Engine 3, or WSE-3. At 72 square inches, it’s the largest commercial chip ever made. The H100, by contrast, is roughly one square inch. Rather than trying to network lots of chips together—an engineering challenge—Cerebras is simply packing all of the power onto a gargantuan semiconductor.

Cerebras’ chip has four trillion transistors, 50 times the computing power of the H100.

Cerebras is bundling the chips into a computing platform called CS-3, which it says can train a large language model like Meta’s Llama in one day, versus one month for Nvidia-based platforms. Cerebras doesn’t sell chips directly. While it’s willing to sell full hardware systems, most of the start-up’s revenue comes from selling access to its systems the same way cloud vendors do. Among its customers: the Mayo Clinic, GSK, and the Lawrence Livermore National Laboratory.

Cerebras CEO Andrew Feldman says the company built eight times as many systems in 2023 as it did the prior year, and it expects the total to increase 10 times in 2024. Cerebras had 2023 revenue of $79 million, and has reached cash flow break-even. The company has raised $715 million in venture capital, and was valued at $4 billion in its most recent round, in 2021.

“The opportunity is to bring supercomputer performance to large enterprises without supercomputer overhead,” Feldman says.

The secret to that, he says, is the dinner-plate-size chips. “Big chips means you don’t have to break up work.”

Jonathan Ross, who played an important role in developing Google’s AI chips, is now running an AI chip start-up called Groq—not to be confused with Elon Musk’s Grok AI model. (All of this grokking stems from a word coined by Robert Heinlein in his 1961 novel, Stranger in a Strange Land; it means to fully understand something in the deepest possible way.)

Like Cerebras, Groq’s strategy is to sell compute time on a consumption basis, rather than selling chips to hardware and cloud companies. Ross says Groq can run popular models like Meta’s Llama 2-70b at 10 times the speed of Nvidia-based systems. (You can try it free on the company’s website.) After Nvidia’s Huang unveiled Blackwell a few weeks ago, Groq cheekily issued a press release that simply said, “Still faster.” The company has announced $367 million in funding to date; Ross says it’s raised additional capital that hasn’t yet been announced.

AI chip start-up d-Matrix is focused on AI inference applications in the data center, leaving the model-building to Nvidia and others. Founded in 2019, d-Matrix uses a design called “in-memory compute,” which CEO Sid Sheth says is an idea that has been around for 30 or 40 years—an approach to speed up computation—but which never had a really good application until AI emerged. D-Matrix, which has raised $160 million, expects to start selling chips next year.

SambaNova, which has raised $1.1 billion from venture investors, is another company taking a systems-based approach to serving the AI market. In addition to chips, its approach aggregates 54 open-source AI models, including those from Google and Meta.

One surprising element of the SambaNova story is that, for many customers, it is installing physical systems in their data centers. “The large majority of enterprise data still sits on premises,” says CEO Rodrigo Liang. He notes that customers come from areas such as banking, healthcare, and government, including the national laboratories at Sandia, Lawrence Livermore, and Los Alamos.

There are plenty of other AI chip start-ups out there. Rain AI, which has seed funding from OpenAI CEO Sam Altman, is focused specifically on energy efficiency, and running large language models on edge devices. Lightmatter is using photonics—light-based technology originally built for quantum computing applications—to improve the speed of networking in AI data centers.

To be sure, these start-ups have big ideas, but they’re still operating at a small scale. It could be years before they’re ready to take on Nvidia. There are potentially big wallets ready to accelerate their progress, though.

Bloomberg has reported that SoftBank Group is considering a $100 billion investment to fund a new AI chip company. (SoftBank is also a SambaNova investor and customer.) Meanwhile, OpenAI’s Altman, according to The Wall Street Journal, has penciled in a plan to invest up to $7 trillion—a figure that seems implausibly large—to build dozens of new chip fabs. Neither SoftBank nor OpenAI would comment on those reports.

The emergence of the growing number of new and potential competition isn’t likely to end Nvidia’s reign as the AI chip champion. But it does mean that Nvidia’s unobstructed domination of a $1 trillion market is coming to an end. Starting now.

Renewable power is used to produce hydrogen, which is stored in underground caverns until it is needed for green energy

On a plain in western Utah, two massive caverns—each roughly big enough to house the Empire State Building—are being hollowed out of rock salt a mile underground.

Salt caverns like these are emerging as one possible solution to the question of how to store solar and wind energy for later use.

It’s a three-step process. First, electricity from solar and wind farms is used to produce hydrogen. Then the hydrogen is stored in caverns like those scheduled to be completed next year at the Advanced Clean Energy Storage project in Delta, Utah. Finally, the hydrogen can be used as a green substitute for climate-warming fossil fuels in uses ranging from power generation to steel manufacture and shipping.

A number of companies in the U.S. and Europe have started to invest in or seriously study salt-cavern projects in the past few years, with government subsidies for clean energy spurring them on.

Cost considerations

Countries around the globe are building huge amounts of wind and solar energy capacity. In the U.S., around 21% of power generated comes from renewable sources now, but the government is aiming for a zero-carbon power grid by 2035, with heavy reliance on renewables to achieve that goal.

The problem is that renewable power generation can fluctuate a lot depending on the time of day or year. Solar-panel output, for instance, stops when the sun sets, and in California can roughly halve in winter versus summer.

Utilities are building big battery installations that can suck up some of that renewable electricity when it’s plentiful during the day, and release it for a few hours in the evening. But the lithium-ion batteries most commonly used today are too small and expensive to absorb the massive amounts of power needed to balance out grids over months or seasons, energy-industry executives say.

Capturing that renewable power by making hydrogen with it and storing the gas underground isn’t cheap. Industry executives say the cost to make a salt cavern could easily exceed $100 million, on top of expenses for the equipment needed to produce the hydrogen. But trying to provide similar storage with batteries is much pricier.

Green Hydrogen International, a company planning a cavern project in South Texas, estimates it would take around 38,500 Tesla Megapacks—a type of battery popular for large-scale utility installations—at an estimated cost of $59 billion to store the amount of energy it is hoping to keep in its caverns, which it estimates will cost $150 million to make.

And just one of the ACES Delta caverns in Utah will be able store more than three times as much energy as all of the utility-scale batteries the U.S. had online at the end of 2023.

Those economics were good enough to attract the interest of Japanese oil-and-gas company Inpex, which is joining GHI in a feasibility study of the Texas project. “What we’re trying to do is take renewables and make them oil-and-gas scale,” says Brian Maxwell, GHI’s chief executive. “You think of (the caverns) as a big underground battery.”

Room to grow

Salt caverns have been used since at least the 1940s to store fossil fuels. The U.S. keeps a good portion of its natural gas underground, as well as its emergency crude-oil reserves, which reside in four huge salt caverns in Texas and Louisiana.

The caverns are typically hollowed out of deposits of rock salt, formed from the remnants of ancient seas that have hardened into layers or been squeezed into pillars or mushroom-shaped domes of salt underground.

Salt deposits have advantages for storing hydrogen, a notoriously tough gas to trap. They are more leakproof than other types of rocks used for storage sites—a feature especially important for hydrogen, which is the smallest molecule in existence. And the rock salt doesn’t react with hydrogen, which can be corrosive to tanks when it is stored above ground.

To create a cavern, engineers drill deep down into a salt deposit, then flush it with massive amounts of water, which slowly erodes the salt and forms a long, tubelike hole, a process that can take two or three years.

Some rock-salt domes in the U.S. are more than a mile in diameter and are capable of housing more than a hundred storage caverns, says Scyller Borglum, vice president for underground storage at engineering firm WSP Global, and deputy project manager for salt-cavern construction at ACES Delta.

That scale will be necessary if hydrogen becomes a significant energy source, versus its current main use in fertilizers and refining. If the U.S. clean-hydrogen market grows to around 10% of the size of the country’s natural-gas market as measured by energy output, more than a thousand new salt caverns will be needed for hydrogen storage, says Mark Shuster, a researcher at the Bureau of Economic Geology at the University of Texas, Austin, who oversees a program studying underground storage for the gas.

Advertisement

Storage starts next year

Michael Ducker, board director of ACES Delta and an executive at a unit of Japan’s Mitsubishi Power, says he got involved in the project in 2019, when the Intermountain Power Agency, owner of a big Utah coal-power plant that supplies electricity to southern California, approached Mitsubishi Power with a plan to replace the plant with one powered by natural gas and hydrogen. The idea was to use the ballooning amount of excess electricity generated by West Coast solar and wind farms—energy that is now unused—to run electrolyzers, equipment that produces hydrogen by separating it out of water. Mitsubishi Power decided to invest in the project and supply turbines for the new plant.

Luckily, the power plant was next to a big salt dome, which already housed caverns for storing natural gas. ACES Delta started drilling for its hydrogen caverns in 2022 and pumping in water last year. It will begin replacing the water in the caverns with hydrogen next year, Ducker says.

When completed, ACES Delta will be the biggest hydrogen-storage site in the world, and one of the few large-scale green hydrogen projects to get off the ground so far. In a sign of the intense interest the project has generated, U.S. energy giant Chevron bought one of the project partners last year, and now holds a majority stake.