Chinese nuclear giant CNNC sets new benchmark in uranium exploration
Fuel is growing in importance as the country seeks to secure supplies used to power reactors and weapons
I’m back from the International Conference on Machine Learning in Vancouver, one of the biggest annual meetups for artificial intelligence researchers. And this year, the conference underscored all the ways in which large language models are still falling short of everyone’s expectations, despite the immense progress that got us to this point.
Researchers even called into question some of the most promising techniques that have gained popularity over the last year, such as chain-of-thought reasoning, or asking the models to describe how they arrived at an answer—their “thoughts,” so to speak.
For instance, one research paper presented at the event explained how chain-of-thought can actually hurt model performance on certain tasks due to “overthinking.” In one example, models were shown strings of letters that followed some rule that the model didn’t know. Then the models were shown another string of letters and asked whether it followed the unknown rule or not.
Humans typically perform better on this task when they’re told to go with their gut feeling. However, the models performed worse when asked to explain their reasoning. Models are great at finding patterns, but when there are so many possibilities for what the unknown rule or pattern might be, they tend to overthink and end up at the wrong answer, the paper argued.
Several papers at ICML pointed out how today’s large language models are rooted in text and that much more work is needed in getting the AI to understand and generate images, videos and audio, otherwise known as multimodality.
One paper from Google DeepMind explained how models that generate audio can only generate tens of seconds of speech at a time before devolving into gibberish because of memory limitations.
The Google DeepMind researchers came up with a new type of model that’s able to generate nearly 20 minutes of speech in one go. Still, it's crazy that even with all the advances in LLMs, speech AI is far behind, especially with the number of startups that want to build AI personal assistants to converse with us.
Another presentation showed how multimodal LLMs fall short when it comes to answering questions where they need to understand images, like diagrams in physics questions.
For instance, in a question that asks the model to predict what a sheet of paper might look like after folding and cutting it in certain ways, the LLM tries to describe (in text form) how it would solve the problem and gets confused. Humans typically arrive at the right answer more easily by visually imagining folding and cutting the paper, according to the research paper.
In another sign of how AI research has gotten more practical and commercial in recent years, ICML also featured lots of papers and presentations about benchmarks, or ways to measure the performance of LLMs on real-world tasks.
These include SWE-Lancer, which measures models’ ability to code by having them complete real tasks on Upwork; WindowsAgentArena, which specifically tests AI agents’ ability to navigate a Windows operating system and apps like Excel and PowerPoint; and LOB-Bench, a benchmark that measures how well AI models can produce certain kinds of financial data. These benchmarks highlight how AI research has gotten more practical and less theoretical in recent years
A Gold Medal in Math
Late Friday night, OpenAI announced that one of its AI models solved problems in this year’s International Math Olympiad at a level that would have earned a human competitor a gold medal, the highest award in the competition.
That is a much-sought-after achievement for AI developers due to the difficulty of the math problems in the competition. The progress could indicate that language models are good enough at picking up mathematical patterns that they will be able to contribute to mathematical research sooner than previously expected.
Last year, Google DeepMind scored a silver medal on the IMO, solving four of the six problems, which first had to be translated into formal mathematical statements that DeepMind’s system could understand.
OpenAI’s model provided correct proofs for five of the six problems, as graded by former IMO competitors that OpenAI hired. Unlike DeepMind’s system, which used a combination of specialized models designed only to solve math problems, OpenAI said it used a general-purpose language model.
Still, not everyone is bowled over by the news. “People are making a big deal over the silver versus gold,” but since there are only six problems, the difference between silver and gold can come down to random noise, said Elliot Glazer, lead mathematician at Epoch AI, which developed the challenging AI math benchmark Frontier Math.
But “it’s more surprising” that OpenAI was able to achieve its score using a general-purpose LLM, he said. Gemini 2.5 Pro, Google’s general purpose LLM, held the previous high score on the IMO among language models, but it did not even earn a bronze medal.
While the IMO problems are difficult, “this is a high school competition,” he said. It’s “many miles from the kind of math that really matters to mathematicians.” In contrast, OpenAI’s o4-mini solves only 6% of the hardest math problems in Frontier Math. Once AI models can solve all those problems, then they will be ready to undertake their own math research, Glazer said.
Another impressive sign of progress came from the performance of OpenAI’s new ChatGPT agent on SpreadsheetBench, a benchmark that tests models’ ability to edit spreadsheets. ChatGPT agent was able to complete up to 45.5% of the tasks correctly, significantly higher than Microsoft’s Copilot product in Excel.—Rocket Drew
Free Software for Content Moderation
Roost, a nonprofit that launched in February to develop content moderation tools, is releasing software for companies, such as productivity and messaging apps, to detect unwanted content online, including AI-generated material.
One of Roost’s tools, Coop, reviews content, such as AI-generated videos, and can route potential child sexual abuse material to the National Center for Missing & Exploited Children. Coop is built atop software developed by Cove, a startup whose intellectual property Roost acquired.
Roost’s other tool, Osprey, helps companies like Discord, the messaging service, monitor and remove posts that violate their rules. Discord developed Osprey before donating it to Roost. Bluesky said it plans to use the product.
“We cannot expect a thousand companies to build a thousand different safety systems,” said Vinay Rao, who joined Roost last week after serving as head of the safeguards team at Anthropic, which detects and patches vulnerabilities with Anthropic’s Claude models.
At Anthropic, users would contact Rao’s team to report potential bugs in Claude. Most of the time, they identified a behavior that Anthropic already knew about, said Rao, but every two weeks or so, someone would disclose a bug that warranted a new defense from Anthropic.
Rao aims to make automatic detection tools like the ones Anthropic uses available to other companies through Roost. He says that even xAI, which has made a point of developing AI that is willing to say provocative things, still needs these kind of trust and safety tools to prevent child sexual abuse material, crypto scams and unwanted bots.
Roost, which is an acronym for Robust Open Online Safety Tools, is releasing the two products open-source, meaning software developers can download them for free and modify them.—Rocket Drew
Scott Bessent calls for probe into ‘the entire Federal Reserve institution’
Treasury secretary is latest top Trump administration official to criticise the US central bank
US Treasury secretary Scott Bessent has called for an inquiry into the “entire Federal Reserve institution”, in the latest sign of how top Trump administration officials are cranking up pressure on the central bank.
“What we need to do is examine the entire Federal Reserve institution and whether they have been successful,” Bessent told CNBC on Monday.
His comments come as President Donald Trump and his lieutenants have sharply criticised the Fed and its chair Jay Powell for refraining from cutting borrowing costs this year. Trump last week asked a group of Republican lawmakers whether he should sack Powell, but later clarified that he had no plans to do so unless he needed to “leave for fraud”.
Bessent later posted on X that while the independence of the Fed “is a cornerstone of continued US economic growth and stability . . . this autonomy is threatened by persistent mandate creep into areas beyond its core mission”.
He added that this provoked “justifiable criticism that unnecessarily casts a cloud over the Fed’s valuable independence on monetary policy”.
The Trump administration has also recently opened a new front in its campaign against the Fed, with the president’s budget director Russell Vought repeatedly alleging that a $2.5bn renovation of the central bank’s headquarters has been grossly mismanaged.
Bessent amped up his criticism of the central bank on Monday, saying that if the Federal Aviation Administration had made as many mistakes, “then we would go back and look at why this has happened”.
The Fed inspector general is reviewing the renovation of its headquarters, which involves an overhaul of two buildings that overlook the National Mall and is $700mn over budget. Powell has also written to senior senators to explain how the US central bank is reining in costs.
On Monday, the Fed published a video tour of the two buildings, which were built in the 1930s, that are undergoing work.
The focus on the plans follows repeated attacks from Trump on Powell over the Fed’s stance on interest rates, with the president labelling the central bank chair a “stubborn mule”.
After enacting 1 percentage point in rate cuts over September to December, most rate-setters want to keep borrowing costs on hold at 4.25-4.5 per cent as they weigh up the impact of the US president’s trade war on inflation.
The Consumer Price Index figure for June showed prices rising at their fastest pace since February, with the headline inflation rate hitting 2.7 per cent. Central bankers have struggled to reach their 2 per cent target since the pandemic, when supply shortages and record fiscal stimulus from the Biden Administration sparked the worst bout of inflation in a generation.
Economists are expecting further rises in price pressures over the coming months, but Trump wants rates slashed to as low as 1 per cent, with the president claiming that the Fed’s policies are raising the government’s refinancing costs by hundreds of billions of dollars.
Bessent claimed that the Fed was “fear-mongering over tariffs” and said the US had seen “great inflation numbers”.
While most investors do not expect the Fed to cut interest rates until at least September, some central bank officials are likely to back lower borrowing costs as soon as next week.
Christopher Waller, a Fed governor who is in the running to replace Powell as chair, has said a weakening of the US labour market justifies a 25 basis points cut in borrowing costs.
Powell is set to step down from his role at the helm of the world’s most powerful central bank in May 2026.
Bessent was seen as another one of the front-runners in the race to lead the Fed. But Trump said this month that he was happy where his Treasury secretary is, before indicating that National Economic Council chair Kevin Hassett was among the top contenders.
How will regulatory overhaul affect water industry in England and Wales?
Cunliffe calls for abolition of regulator Ofwat but steers clear of proposing nationalisation of the embattled sector
The death warrant for water regulator Ofwat appeared in a 454-page report published on Monday by the Independent Water Commission, after it pored over 50,000 responses and met almost 250 organisations and individuals.
Sir Jon Cunliffe, the former Bank of England deputy governor who chaired the review, was ordered by ministers to rip up the current system of oversight for water companies in England and Wales — but stop short of nationalisation.
His report comes against a backdrop of public anger over the industry’s generous pay packets, high levels of sewage pollution and rising household bills.
What will be the immediate consequences of the report?
The commission called for Ofwat to be abolished, with a new regulator also replacing the Drinking Water Inspectorate while assuming the water-related functions of the Environment Agency and Natural England.
In Wales, Ofwat’s economic responsibilities should be integrated into Natural Resources Wales, an existing body, Cunliffe said.
The report warned that the regulatory landscape for water was far too fragmented and said “fully joined-up regulation is essential for the system to meet the demands of the future”.
Speaking after the review was published, environment secretary Steve Reed promised to legislate quickly to introduce a new regulator.
Cunliffe, who gave a separate speech on Monday, said Ofwat had been encouraged to monitor the sector with a light touch and keep bills down. It had also lacked the powers or capability to supervise the industry’s complex financial structure, he added.
This had “allowed some companies and their owners to take decisions which reflected their private interests but badly damaged both their companies and in the longer term the public interest”, he said.
The report did not name individual companies, but Australian investor Macquarie increased Thames Water’s debt from £3.4bn in 2006 to £10.8bn in 2017 while extracting billions in dividends and loans.
Reed has so far accepted four other recommendations from Cunliffe.
They include strengthening the legal powers of the water ombudsman, the Consumer Council for Water, and reforming self-monitoring of wastewater compliance by companies.
Most of Cunliffe’s recommendations will be considered by ministers at leisure.
Has Ofwat’s demise been welcomed?
Surfers Against Sewage, a campaign group, described the proposals to axe the regulator as “putting lipstick on a pig”, given that the industry’s financial and ownership structures will remain intact.
But Water UK, the industry trade body, said the report offered “fundamental change” and a “significant step forward”.
“These recommendations should establish the foundations to secure our water supplies, support economic growth and end sewage entering our rivers and seas,” it said.
What about nationalisation?
Nationalisation was deliberately kept outside the inquiry’s scope. Reed said it would cost taxpayers about £100bn that could otherwise be spent on schools and hospitals.
Cunliffe insisted that Britain had been the “dirty man of Europe” before then prime minister Margaret Thatcher’s 1989 water privatisation and that the sell-off had been a success at first. Only in recent years had the performance of the industry “plateaued”, he said.
What is the proposed turnaround regime?
Cunliffe argued against “punitive measures on poorly performing companies that make it harder for them to recover”.
The management of heavily indebted Thames Water, which provides water and sewerage services to about a quarter of the UK population, has been begging for a more lenient approach to pollution spills to help it survive.
The commission proposed the regulator should have a new turnaround regime to support the recovery of poorly performing companies.
This could allow the regulator to “defer or waive fines” if they damaged a utility’s ability to make infrastructure investments. Restrictions on management bonuses and shareholder dividends were mooted as a condition of entering the regime.
What will the shake-up mean for water bills?
Reed vowed water customers would “never again be hit with unaffordable rises in their bills”.
In December, Ofwat announced that water companies could raise customer bills by an average of 36 per cent over five years to 2030.
When ministers boast about £104bn of new investment for the water sector, they gloss over the fact this is being funded by those higher bills.
The Cunliffe report suggested bills were previously held down between 2009 and 2024 at the behest of politicians and the regulator, throttling much-needed investment in the network.
Reed’s argument is that if water companies had paid less in dividends and bonuses, they could have invested more in infrastructure upgrades.
But Cunliffe was sanguine that bills needed to rise to provide necessary investment.
“We have to recognise that the cost of producing water and waste water services is likely to increase over the medium and longer term,” he said.
Will there be tighter restraint on executive pay?
In addition to sewage spills, there has been public anger about high pay packets at underperforming water companies.
Ofwat has new powers to restrict bonuses for water bosses, but only six have been banned this year.
By contrast, Thames alone was able to pay bonuses worth £2.5mn in April to 21 senior staff as a reward for taking on a punitively high-interest loan. Meanwhile Southern Water has sidestepped the ban by doubling its chief executive’s overall packet.
Yet Cunliffe’s report shied away from tougher restrictions on water executives’ pay.
Ministers prepare to reduce UK stake in Sizewell C nuclear project
Under changes to investors, Brookfield dropped in favour of Canadian pension fund and London-based investment manager
Ministers are preparing to reduce the UK government’s stake in Sizewell C by more than expected, after a last-minute move to change the roster of private investors in the multibillion-pound nuclear project.
Canadian investment manager Brookfield had been expected to take a 25 per cent stake in the Suffolk development as recently as Friday, according to people familiar with the matter.
The proposed deal would have made the firm the biggest single investor in the project — part of a push by ministers to use nuclear power to ease Britain’s transition away from fossil fuels — with British Gas owner Centrica taking 15 per cent and French energy giant EDF holding 12.5 per cent.
But the government in recent days moved to drop Brookfield in favour of selling a larger stake to a combination of Canadian pension fund La Caisse and London-based investment manager Amber Infrastructure, the people said.
Two of the people said Brookfield had been angered by the decision, which was first reported by Les Echos.
La Caisse will now take a 20 per cent stake in Sizewell and Amber will take a 10 per cent stake, according to the people, while the stakes of Centrica and EDF are unchanged.
The increased private investor stake means the UK government will own 42.5 per cent of the project, which is set to cost £38bn, down from 47.5 per cent under previous plans. The government currently owns about 84 per cent of the project.
Brookfield declined to comment, as did La Caisse, Amber Infrastructure and the government.
Charles Emond, chief executive of La Caisse, which manages C$473bn (£254bn) on behalf of 6mn pension savers, told the Financial Times in May that it planned to invest more than £8bn in the UK over the next five years.
“We’d like to be a partner of trust and choice in the UK,” said Emond. The government’s plans to increase infrastructure spending were “a huge opportunity and we’d like to be there in the early stages to see if we can do something”, he added.
UK energy secretary Ed Miliband said last month that Sizewell would be the beginning of a “golden age” for nuclear in Britain. He also said the project would be “majority public funded”. The government has committed £17.8bn.
Ministers are set to announce the final investment decision on Tuesday in a rush to seal the long-awaited deal before parliament’s summer recess.
Energy bill payers will pay back the investors via a surcharge on bills under the structure used to develop the scheme, known as the “Regulated Asset Base” model.
Under the RAB model, as used in the £4.2bn Thames Tideway sewer project, investors start getting paid during construction, rather than having to wait until the project is finished. The aim of this is to reduce risks and therefore financing costs for the project.
AstraZeneca Plans to Invest $50 Billion in U.S. by 2030
Plans include a proposal for a new drug substance manufacturing center focused on weight-loss drugs
- AstraZeneca plans to invest $50 billion in the U.S. by 2030, including a new drug substance manufacturing center.
- The Virginia facility would be AstraZeneca’s largest single manufacturing investment in the world.
- The investment aims to boost research-and-development, expanding facilities in Maryland and Massachusetts.
AstraZeneca AZN -0.60%decrease; red down pointing triangle said it plans to invest $50 billion in the U.S. by 2030, marking the latest drugmaker to commit to American manufacturing as looming tariffs threaten to raise costs across the industry.
The biopharmaceutical giant said Monday the investment includes plans for a new drug substance manufacturing center focused on chronic diseases. The multibillion-dollar facility would produce drug substances for the company’s weight-management and metabolic portfolio, including oral GLP-1s.
The Virginia facility would be AstraZeneca’s largest single manufacturing investment in the world, the company said.
Other drugmakers such as Eli Lilly LLY -1.23%decrease; red down pointing triangle & Co., Roche Holding ROG -1.12%decrease; red down pointing triangle, Regeneron Pharmaceuticals REGN 0.55%increase; green up pointing triangle and Merck MRK -0.81%decrease; red down pointing triangle & Co. have similarly committed billions to expand their U.S. manufacturing efforts in recent months.
The planned investments come as President Trump threatens to impose tariffs of up to 200% on pharmaceuticals made abroad. Companies would be given a runway of up to a year and a half to reshore supply chains before the levies come into effect, Trump said earlier this month.
Commerce Secretary Howard Lutnick said Trump’s new tariff policies are focused on ending the U.S.’s reliance on foreign supply of key pharmaceutical products. “We are proud that AstraZeneca has made the decision to bring substantial pharmaceutical production to our shores,” Lutnick said.
The U.K. company’s planned investment also aims to prop up its research-and-development efforts, expanding a facility in Gaithersburg, Md., and building a new center in Cambridge, Mass.
It would also add manufacturing facilities for cell therapy in Rockville, Md., and Tarzana, Calif., and expand manufacturing in Mount Vernon, Ind., and Coppell, Texas, AstraZeneca said.
Chief Executive Pascal Soriot said the investment would support AstraZeneca’s goal of $80 billion in revenue by 2030. It is also expected to create tens of thousands of new jobs across U.S., the company said.
Universal Music Group Files for Public Offering
Netherlands-based music conglomerate says it has filed a confidential registration statement with the SEC proposing the offering
- Universal Music Group confidentially filed for a U.S. public offering with the Securities and Exchange Commission.
- The music conglomerate will not receive proceeds from the sale of ordinary shares by the selling shareholders.
- Pershing Square can request UMG to list in the U.S. if it sells at least $500 million in stock as part of the offering.
Universal Music Group UMG 0.30%increase; green up pointing triangle filed for a public offering in the U.S.
The Netherlands-based music conglomerate said Monday it had filed a confidential registration statement with the Securities and Exchange Commission proposing the offering.
UMG won’t receive any proceeds from the sale of ordinary shares by the selling shareholders, it said. The number of shares and price range for the offering haven’t yet been determined.
UMG, which owns the record label behind Taylor Swift and Lady Gaga, trades on the Euronext Amsterdam exchange and is up 10% this year.
UMG and Bill Ackman’s Pershing Square have an agreement that Pershing Square can request UMG to list in the U.S. if it sells at least $500 million in stock as part of the offering. In January, UMG said Pershing Square had exercised its right and the music company aimed to carry out an offering of certain Pershing-owned shares by September. Ackman resigned from his role on UMG’s board of directors in May.
SoftBank and OpenAI’s $500 Billion AI Project Struggles to Get Off Ground
The Stargate venture, introduced at White House event, is now setting the more modest goal of building a small data center by year-end
The $500 billion Stargate project, intended to boost U.S. AI, faces setbacks six months after its White House unveiling.
SoftBank and OpenAI are in disagreement over key partnership terms, leading to a scaled-back plan for a data center in Ohio.
Despite Stargate’s slow progress, OpenAI is pursuing data-center deals independently, including a $30 billion annual deal with Oracle.
A $500 billion effort unveiled at the White House to supercharge the U.S.’s artificial-intelligence ambitions has struggled to get off the ground and has sharply scaled back its near-term plans.
Six months after Japanese billionaire Masayoshi Son stood shoulder to shoulder with Sam Altman and President Trump to announce the Stargate project, the newly formed company charged with making it happen has yet to complete a single deal for a data center.
Son’s SoftBank 9434 0.09%increase; green up pointing triangle and Altman’s OpenAI, which jointly lead Stargate, have been at odds over crucial terms of the partnership, including where to build the sites, according to people familiar with the matter.
While the companies pledged at the January announcement to invest $100 billion “immediately,” the project is now setting the more modest goal of building a small data center by the end of this year, likely in Ohio, the people said.
Stargate’s lethargic launch is a setback to the vast ambitions of Son, who despite spending billions of dollars over the years, has been playing catch-up in the fast-evolving AI sector.
SoftBank committed $30 billion to OpenAI earlier this year. It is by far the largest-ever startup investment—an enormous wager that has led SoftBank to take on new debt and sell assets. The investment was made alongside the plans for Stargate, giving SoftBank a role in the physical infrastructure needed for AI.
Altman, eager to secure the computing power to support the next generations of his company’s signature product, ChatGPT, has plowed ahead without SoftBank, signing deals for data centers with other operators.
The leaders of both companies say all is well in their joint effort. Last week they appeared on video at a SoftBank event, and Altman said they have an initial goal of building 10 gigawatts of data centers together. It is a “wonderful partnership,” he said.
In a joint statement, the two companies said they were advancing projects in multiple states and were “moving at hyperscale and speed to deliver the AI infrastructure that will power the future and serve humanity.”
Stargate’s rocky start hasn’t slowed the data-center development spree that Trump has said is a national priority. AI enthusiasts say the world will require a gargantuan effort to build the warehouse-like structures filled with computer servers—and the electricity needed to supply them—on par with the construction of railroads in the 19th century.
Altman’s OpenAI recently struck a data-center deal with Oracle that calls for OpenAI to pay more than $30 billion a year to the software and cloud-computing company starting within three years, according to people familiar with the transaction.
That deal, which doesn’t involve SoftBank, totals 4.5 gigawatts of capacity, and would consume the equivalent power of more than two Hoover Dams, enough to power about four million homes. The data centers are spread among locations around the U.S., people familiar with the deal said.
Taken with a smaller deal OpenAI struck with CoreWeave, OpenAI has now completed data-center deals for nearly as much capacity as Stargate promised for this year in January. (OpenAI has said $100 billion roughly equates to 5 gigawatts of data centers.)
Despite Stargate’s slow start, Son has told associates he is bullish on OpenAI and would like to invest even more in the company, a person familiar with the matter said.
Son has long craved a prominent seat at the AI table. Over the past decade, he devoted the two largest startup funds ever raised—more than $140 billion—to finding the AI companies of the future, only to miss out on OpenAI and all of its competitors before the launch of ChatGPT, while being tarred by high-profile flops such as WeWork and construction startup Katerra.
He had better success with his 2016 acquisition of chip company Arm, which surged in value during the past two years.
In the fall, he zeroed in on OpenAI. Altman had ambitions on a scale without precedent: He has said trillions of dollars will be needed to fund the next stages of AI.
In November, Son and Altman met in Tokyo to hash out an agreement. They next brought their plans to Trump aides with the offer of announcing a giant U.S. investment. On the second day of Trump’s new administration, they stood smiling in the White House beside Trump, vowing to spend $500 billion by 2029.
“This is the beginning of our golden age,” Son said at the White House.
OpenAI and SoftBank each pledged $18 billion to the venture to get started, which they expected would develop data centers and then lease them to OpenAI.
The two companies said they would jointly control the entity, with Son as its chairman. SoftBank would focus on the financial details and OpenAI on the operations, the companies added.
They named Oracle and U.A.E. firm MGX as partners in the initial announcement, but their role was unclear. No agreement setting out their investment levels had been reached at the time.
“Stargate is not formed yet,” Oracle Chief Executive Safra Catz said on an investor call last month.
One recent complication between OpenAI and SoftBank has been over how extensively to build data centers on sites tied to SB Energy, a SoftBank-backed energy developer, according to people familiar with the matter.
Altman has used the Stargate name, shared with a 1994 Kurt Russell film about aliens who teleport to ancient Egypt, on projects that aren’t being financed by the partnership between OpenAI and SoftBank. The trademark to Stargate is held by SoftBank, according to public filings.
For instance, OpenAI refers to a data center in Abilene, Texas, and another it agreed in March to use in Denton, Texas, as part of Stargate even though they are being done without SoftBank, some of the people familiar with the matter said.
Building data centers isn’t easy. Companies need to find sites, develop the physical structures—or pay another company to do that—buy expensive AI chips, source vast loads of electricity and find lenders to cover much of the cost, among other details. Stargate is hoping to use a new, lower-cost design for its first project in Ohio, people familiar with the matter said.
Training new, better AI models requires evermore data processing, which means companies including OpenAI are committing to larger and larger spending well before the business models of AI companies have proved profitable.
OpenAI’s $30-billion-a-year deal with Oracle is roughly three times as much as what Altman’s company has recently projected in annual revenue. The company, which is losing billions of dollars a year, is betting its revenue will multiply thanks to the growing ranks of paying customers and advertising, allowing it to pay for its hefty commitments.
After Hours Summary: MEDP +48.9%, CALX +4.2%, WTFC +2.8% higher on earnings; AGYS -10.6%, NXPI -5.4%, STLD -1.7% lower on earnings
After Hours Gainers:
Companies trading higher in after hours in reaction to earnings/guidance: MEDP +48.9%, CALX +4.2%, WTFC +2.8%, ZION +1.6%, ELS +1.3%, CCK +0.2%
Companies trading higher in after hours in reaction to news: CSX +4.7% (CP and CSX form new rail routing option called Southeast Mexico Express), NFGC +3.3% (assessment for Queensway gold project), PARR +2.5% (establish joint venture for renewable fuels in Hawaii), ASH +1.8% (names new CFO), GD +0.5% (awarded a $159.7 mln modification to a previously awarded US Navy contract), LMT +0.4% (awarded a $999 mln US Air Force contract)
After Hours Losers:
Companies trading lower in after hours in reaction to earnings/guidance: AGYS -10.6%, NXPI -5.4%, STLD -1.7%, WRB -1.5%, BOKF -1.4%, ARE -0.1%
Companies trading lower in after hours in reaction to news: KRMN -4% (files for 20 mln share offering by selling shareholders; also provides guidance), SFNC -3.7% (files prospectus supplement for $250 mln stock offering), ORCL -2.3% (Softbank and OpenAI's $500 bln AI project scaled back near term, according to WSJ), ECX -0.8% (files for $255 mln mixed securities shelf offering), EQH -0.7% (files mixed securities shelf offering), CP -0.3% (CP and CSX form new rail routing option called Southeast Mexico Express), NBHC -0.1% (strategic partnership to support 2UniFi), BRT -0.1% (acquires Alabama property)