The Information : OpenAI Builds an M&A War Chest

OpenAI Builds an M&A War Chest

Will OpenAI make more acquisitions? The ChatGPT maker is in the process of raising $100 billion in additional cash. I wouldn’t be surprised if it uses some of that money or its richly valued shares to buy startups or lure even more talent.

After all, it’s done this before. It used its stock to buy hardware guru Jony Ive’s startup, Io Products, in a $6.5 billion deal, when OpenAI was in the middle of raising a round that valued it at $260 billion before the funding. OpenAI is now on the cusp of a financing that will value it at more than $730 billion before the new capital—a higher market cap than all but the top 15 public companies.

Of course, OpenAI will need to use the bulk of its latest haul to support its fast-growing, cash-burning business. It has projected it will spend a whopping $665 billion just on computing costs over the next five years, as my colleagues have reported.

Against those costs, a startup valued at a few billion dollars looks like pocket change. And thanks to a corporate restructuring last fall, OpenAI can also court founders with traditional equity rather than less common profit-sharing units.

What would it buy? It may go after talented engineers, as it did recently by hiring Peter Steinberger, founder of the popular open-source personal agent project OpenClaw. It could buy startups whose staff have expertise in fields where it’s trying to expand, such as healthcare or enterprise software. Earlier this year, for instance, it agreed to pay $100 million in stock, including equity to retain employees, for healthcare app Torch.

OpenAI has also shown an interest in buying a coding-related business. Last year, it agreed to buy Windsurf for $3 billion before the deal fell apart and the founders went to Google. Around that time, it also held conversations to buy Cursor, the best known of the coding assistant startups.

OpenAI could revisit its Cursor talks or buy one of several other coding-related businesses, as it seeks to catch up to what Anthropic’s Claude Code has achieved in automating coding. Cursor, last valued at $30 billion, would be OpenAI’s largest acquisition to date if it happened. But it would only be about 4% of the company’s new valuation!

We know OpenAI is hungry for both people and data, the lifeblood of AI development. Earlier this year, we reported that OpenAI, Anthropic and Google have all been on the prowl for specialized data to train their AI models. These could turn into licensing deals with life sciences or financial services companies—or even outright acquisitions.

I asked ChatGPT what OpenAI could acquire, and it said, “The logic would likely fall into a few buckets: distribution, proprietary data, hardware, vertical AI apps and infrastructure.”

That’s pretty broad! Either OpenAI has more mergers and acquisitions in the works—or its chatbot is hallucinating.

The Information : Meta’s Internal Chip Design Efforts Hit Roadblocks

Meta’s Internal Chip Design Efforts Hit Roadblocks

The Takeaway
  • Meta scraps advanced AI training chip Olympus due to design struggles.
  • Decision underscores difficulty competing with Nvidia’s dominant AI chips.
  • Meta has struck chip supply deals with AMD and Nvidia for data centers.

As Meta Platforms strikes new chip supply deals with AMD and Nvidia, it has been running into problems with AI chips it is designing internally, according to six people with direct knowledge of the matter.

Meta last week scrapped the most advanced chip it was developing for training AI models, after struggling with the chip’s design, and shifted its focus to a less complicated version, the people said. The company informed staffers in its AI infrastructure division about the updated plans last week. The decision underscores the difficulty tech giants face in designing AI chips that can compete with Nvidia’s offerings, which dominate the market.

Meta’s revised chip road map follows new partnerships it has struck with Advanced Micro Devices and Nvidia in recent weeks. Meta and AMD announced on Tuesday that the social media giant will purchase 6 gigawatts’ worth of AMD’s chips—roughly equivalent to the power needed to run several large-scale data centers. Meta also announced a “multigenerational” partnership with Nvidia earlier this month, in which it is committing to deploy millions of Nvidia’s current and next-generation chips in its data centers.

Meta’s in-house AI chips, which fall under its Meta Training and Inference Accelerator program, are part of the company’s broader effort to develop its own AI hardware and reduce its reliance on outside chipmakers like Nvidia. The move aims to cut costs and improve control over its data center infrastructure.

Meta, for example, expects its capital expenditures in 2026 to range from $115 billion to $135 billion. Most of this spending will go toward investments in chips and servers.

In a statement, a Meta spokesperson said: “We remain committed to investing in a diverse silicon portfolio to meet our needs, which includes advancing our MTIA portfolio and will have more to share this year.”

Other companies such as Microsoft are encountering similar issues designing their own AI chips. Last year, Nvidia CEO Jensen Huang said publicly that most big tech firms would abandon the rival chip projects they are pursuing. He predicted that the performance of those chips would continuously fall short of Nvidia’s chips.

Meta has encountered problems with several of its own chips. The company scrapped one version of its second-generation training chip, internally code-named Iris. It then began working on a more advanced training chip, code-named Olympus, but it has now scrapped that one as well.

One person who works on Meta’s chips said there is internal skepticism about the company’s attempt to build chips that match Nvidia’s in capabilities given the risk of delays or redesigns. Such a task requires a large team of engineers to design and debug the chips and to ensure their power consumption isn’t excessively high, which wouldn’t make the chips worth using versus Nvidia’s offerings, the person said.

The Iris training chip is based on a computing approach known as single instruction, multiple data. SIMD is generally easier for hardware engineers to design but harder for software engineers to program when training AI models, the people said.

Olympus was based on a computing approach similar to that used for Nvidia’s AI chips: Single instruction, multiple threads (SIMT) makes it generally easier for software engineers to program but harder for hardware engineers to design.

Many tech companies favor this approach, which Nvidia popularized, as it offers more flexibility and is better suited for training modern AI models.

Meta intended to complete the design of Olympus by the fourth quarter of 2026 at the earliest, though new chip designs typically take an additional nine months or longer beyond initial development to mass-produce, according to four of the people. The central part of Olympus that handled AI calculations—the graphics processing unit—would have utilized a design from chip startup Rivos, acquired by Meta last year.

Rivos said its GPUs could efficiently run Nvidia’s proprietary Cuda software code, the dominant software for training and running machine-learning models.

Meta initially planned to build large clusters of servers with Olympus, but executives ultimately decided that doing so would have posed a major risk to training new models as it races to compete against more established rivals such as OpenAI and Google, one of the people said. The software for training the chips wouldn’t have been as stable as Nvidia’s offerings, for example, and Olympus’ complicated design could have made it difficult to manufacture in large quantities, several people said.

Instead, Meta is opting for now to continue using training chips made by others, for which the software is more established, the person said.

>>> Netflix declines to raise offer for Warner Bros (WBD) (84.59 +1.88)

Netflix declines to raise offer for Warner Bros (WBD) (84.59 +1.88)
  • Co announces that it has declined to raise its offer for Warner Bros. Netflix had earlier received notice from Warner Bros. Discovery (WBD) that its Board of Directors has determined Paramount Skydance's (PSKY) latest proposal constitutes a "Superior Proposal" under the terms of WBD's existing merger agreement with Netflix.
  • Co issued the following statement in response from co-CEOs Ted Sarandos and Greg Peters:
    • The transaction we negotiated would have created shareholder value with a clear path to regulatory approval. However, we've always been disciplined, and at the price required to match Paramount Skydance's latest offer, the deal is no longer financially attractive, so we are declining to match the Paramount Skydance bid.
    • Warner Bros. is a world-class organization, and we want to thank David Zaslav, Gunnar Wiedenfels, Bruce Campbell, Brad Singer and the WBD Board for running a fair and rigorous process. We believe we would have been strong stewards of Warner Bros.' iconic brands, and that our deal would have strengthened the entertainment industry and preserved and created more production jobs in the U.S. But this transaction was always a 'nice to have' at the right price, not a 'must have' at any price.
    • Netflix's business is healthy, strong and growing organically, powered by our slate and best-in-class streaming service. This year, we'll invest approximately $20 billion in quality films and series and will expand our entertaining offering. Consistent with our capital allocation policy, we'll also resume our share repurchase program.
    • We will continue to do what we've done for more than 20 years as a public company: delight our members, profitably grow our business, and drive long-term shareholder value.

The Information : Google Strikes Multibillion-Dollar AI Chip Deal With Meta, Sha

Google Strikes Multibillion-Dollar AI Chip Deal With Meta, Sharpening Nvidia Rivalry

The Takeaway
  • Google accelerates TPU sales, challenging Nvidia’s market dominance.
  • Google forms joint venture to lease TPUs to other AI customers.
  • Google and Meta are still among the biggest buyers of Nvidia AI chips.

Meta Platforms has signed a deal to rent Google’s AI chips, known as tensor processing units, to develop new AI models, according to a person involved in the talks. The multi-year deal is worth billions of dollars, said a person who was briefed about it. Meta has also been talking to Google about buying TPUs for its data centers as soon as next year, though the status of that discussion couldn’t be learned.

The deal is a victory for Google, giving it another brand-name customer that could help its efforts to build a multibillion-dollar business selling its TPUs. At the same time, it poses a threat to Nvidia, which dominates the AI chip market and currently supplies Meta with the graphics processing units to develop its AI, a process known as training.

In addition to forging the Meta deal, Google has signed an agreement with an unidentified large investment firm to fund a joint venture that would lease TPUs to other customers, according to a person involved in that arrangement. Google is in talks with other investment firms to fund other such joint ventures.

Both moves show Google is accelerating efforts to compete directly with Nvidia in the AI chip business, including in the AI training market that Nvidia dominates. Google stands to generate billions in additional revenue from selling TPUs.

News of the Google-Meta deal comes days after Nvidia announced a new deal with Meta, in which the Facebook owner said it will buy millions of GPUs for its data centers in the coming years. At the time, the Nvidia deal raised questions about the future of Google’s talks with Meta, which The Information reported last year, but it isn’t clear that it had any impact. Another factor in the Meta-TPU deal could be Meta’s own struggles to develop and AI training chip, details of which The Information reported earlier Thursday.

Some leaders of the Google Cloud unit previously suggested internally that supercharging the TPU business could help the company grab as much as 10% of Nvidia’s annual revenue—around $200 billion in the past 12 months, according to a person who heard the remarks.

(For details on how TPUs compare to Nvidia GPUs, see this article.)

Google is looking at a variety of ways to get TPUs into the hands of customers, according to a person who has been involved in its TPU strategy. That’s why the company is in talks with private equity firms about launching joint ventures that would buy its TPUs and then lease them to AI customers, according to two people with direct knowledge of these conversations.

The joint ventures could also start cloud businesses and be responsible for operating the TPUs. Nvidia has taken similar steps to seed customers for its chips, including funding a wave of upstart cloud providers, known as neoclouds, that focus on renting out its GPUs to AI customers.

Google has signed at least one term sheet with a large investment firm as part of this effort, one of these people said. A small group led by Google Cloud veteran Benjamin Treynor Sloss is leading the TPU financing effort, working closely with cloud chief Thomas Kurian, one of the people said.

Google’s corporate development team is also meeting with potential finance partners to borrow money for special purpose vehicles that would buy TPUs and lease them to customers, according to someone involved in those conversations. That would look similar to how Elon Musk’s xAI has structured creative financing deals with venture capital firm Valor to get access to Nvidia GPUs. In the deals under discussion with Google, TPUs could be used as collateral for the debt, the person said.

Balancing Act

Managing the growth of the TPU business is a tricky balancing act for Google. While the tech giant is increasingly competing with Nvidia, its Google Cloud unit is also one of Nvidia’s biggest GPU customers. That’s because most AI developer customers want GPUs to develop their technology, and Google Cloud can’t afford not to offer Nvidia-powered servers. Google will need a continuous supply of Nvidia’s latest chips to stay competitive in the cloud market.

And the number of TPUs available to new customers is also a question mark. Google’s own AI team, which develops the Gemini chatbot and models, uses TPUs to develop its technology and must ensure it has enough supply to compete with the likes of OpenAI. Taiwan Semiconductor Manufacturing Co. produces both TPUs and Nvidia GPUs, meaning they effectively vie for capacity at its facilities.

Google for years has rented out TPUs to cloud customers who use them in Google Cloud data centers, but last year it began pitching some of those customers—including Meta and big financial institutions—on the idea of using TPUs in their own data centers.

One of the ways it is doing so is by pitching the cost benefit of TPUs, arguing they’re cheaper to use than pricey Nvidia chips. The high prices for state-of-the-art Nvidia chips have also made it difficult for other cloud providers, such as Oracle, to generate solid gross profit margins from renting out Nvidia chips to AI developers.

Google’s TPU push also comes after some cloud providers and customers, including OpenAI and Meta, struggled last year to get Nvidia’s latest Blackwell AI chips up and running at the scale they wanted, partly due to technical glitches and other complexities involving the hardware.

Meta and other companies that directly purchase AI chips for their own data centers have long sought alternatives to Nvidia so they aren’t beholden to one supplier. Meta this week announced a large deal to buy AI chips from another Nvidia competitor, Advanced Micro Devices, though Meta would primarily use those chips to run its existing AI models, a process known as inference, rather than train new ones, said a person involved in the deal talks.

Indeed, the fact that Meta plans to use TPUs for AI training is notable because most analysts, doubting that anyone could compete technically with Nvidia on training, have said they believe the biggest opportunities for challenging Nvidia lie in using chips for inference, which doesn’t require large, interconnected clusters of servers.

Meta is also continuing to develop its own chips for AI inference to save on costs and diversify away from Nvidia’s chips.

Meta won’t be the first large customer of TPUs. Anthropic last year agreed to spend about $20 billion buying TPUs from Broadcom, which co-designs the chips with Google and oversees TSMC’s manufacturing of them. Anthropic plans to use the chips in data centers not run by Google, according to people who have spoken to Anthropic leaders.

TPU-Trained Models

Anthropic for years has used TPUs to both develop and run its Claude AI, and Google has invested billions of dollars in the startup. The Information last summer reported that Google was discussing a TPU deal with OpenAI, and in November it reported the deal talks between Meta and Google over TPUs.

Meanwhile, it’s not lost on Nvidia CEO Jensen Huang that two of the world’s best AI models, from Google and Anthropic, were developed fully or partly using AI server chips made by Google rather than Nvidia GPUs.

He has closely monitored Google’s technical progress with TPUs and has sprung into action to entice existing and potential TPU customers to make big commitments to Nvidia’s GPUs. Meta and Nvidia recently said they negotiated a new partnership, and last year Nvidia made an investment in Anthropic, and got a commitment from Anthropic to use its chips. Nvidia also has discussed making a $30 billion equity investment in OpenAI, The Information reported.

Those moves show that Huang and Nvidia are resisting Google’s efforts to expand its presence in the AI chip market.

Nvidia’s domination of the market for AI server chips has turned it into the world’s most valuable company, with a $4.8 trillion market capitalization. Its stranglehold on the market has also sent its revenues soaring, lifting the amount of cash it throws off—which it can in turn use to invest in customers such as OpenAI and Anthropic, as well as cloud providers like CoreWeave that rent out GPUs.

>>> US After Hours Summary: XYZ +22.9%, AAOI +17.2%, DELL +10.3% higher on earni

After Hours Summary: XYZ +22.9%, AAOI +17.2%, DELL +10.3% higher on earnings; NFLX +10.6% and PSKY +5.3% higher as NFLX declines to match PSKY higher offer; DUOL -21.4%, ZS -9.6%, ESTC -9%, CRWV -8.6%, FLUT -8.6% lower on earnings

After Hours Gainers:

Companies trading higher in after hours in reaction to earnings/guidance: XYZ +22.9% (also plans 40% workforce cut with $450--$500 mln restructuring charges), AAOI +17.2%, CAI +16.8% (also completes interim readout of achieve 1 Study), MARA +16.2% (also strategic agreement with Starwood Capital), PRAA +15.3%, ARLO +13%, EVTC +11.6% (also increases share repurchase authorization to $150 mln), NATL +10.5% (also to be acquired by BCO), REAL +10.4%, DELL +10.3% (also increases dividend by 20%; also $10 bln increase in share repurchase authorization), DRH +8.3%, RKT +8.3% (also announces strategic alliance with COMP), ASUR +7.7%, XRAY +7% (also restructuring plan; eliminates dividend), PUBM +6.6%, BLFS +6.5%, FIGS +6.2%, CON +5.8%, CSTL +5.5%, BWIN +4.6% (also authorizes new $250 mln share repurchase program), AVPT +4.1%, RLAY +3.6%, DEC +2.8% (also acquries natural gas properties), CRNX +2.7%, ARX +2.7%, FOXF +2.4%, RLJ +2.3%, NTAP +1.9%, GLOB +1.7%, MTZ +1.7%, GRND +1.6%, STRA +1.6%, PBA +1.1%, NEXA +1.1%, OPK +0.8%, ADSK +0.7%, CTRA +0.6%, MAIN +0.5%, RCKT +0.4%, MOH +0.2%, NTRA +0.1%, NTRA +0.1%

Companies trading higher in after hours in reaction to news: NGNE +24.5% (FDA grants Breakthrough Therapy designation to NGN-401), NFLX +10.6% (WBD determines revised proposal from PSKY constitutes a superior proposal; NFLX declines to match offer), IMDX +8.1% (enters into specimen collection agreement with DGX), PSKY +5.3% (WBD determines revised proposal from PSKY constitutes a superior proposal; NFLX declines to match offer), AEVA +1.1% (top 5 OEM has selected Aeva for its vehicle development program, also reports earnings), EPSN +1% (authorizes new 10% share repurchase program), AXS +0.9% (authorizes new $300 mln share repurchase program), NTR +0.4% (announces TSX approval for its renewed share repurchase program), AIR +0.3% (announces deal with Otto Instrument), SU +0.3% (renews NCIB to repurchase up to 118.7 mln shares), GOOG +0.2% (Google strikes AI chip deal with Meta, which sharpens Nvidia rivalry, according to TheInfo), XERS +0.1% (file Recorlev patent infringement lawsuit), NVDA +0.1% (Google strikes AI chip deal with Meta, which sharpens Nvidia rivalry, according to TheInfo)

After Hours Losers:

Companies trading lower in after hours in reaction to earnings/guidance: XPOF -21.7%, DUOL -21.4% (also authorizes new $400 mln share repurchase program), EBS -19.6% (also announces series of new contracts; also authorizes new $50 mln share repurchase program), DNA -18.9% (also announces focus on autonomous labs offerings and divestiture of its non-core biosecurity business), WLDN -13.1%, SEER -13%, AMBA -11.1%, ZS -9.6%, ESTC -9%, CRWV -8.6%, FLUT -8.6%, PSNL -8.1%, FIGR -7.1% (also partnership with Agora Data; also authorizes new $200 mln share repurchase program), PAR -6.5% (also authorizes new $100 mln share repurchase program), BCO -6.3% (also to acquire NATL), SMR -6.1%, SMR -6.1%, SG -6%, PCRX -5.7%, ALHC -5.6%, AUGO -5.6% (also to relocate the federal road which will convert resources into probable mineral reserves), PGNY -5.6%, BWMX -5.6%, LASR -5.5%, INOD -5.4%, PBYI -5.4%, INTU -4.9% (also increases dividend), RKLB -4.9% (also acquires Optical Support and acquires Precision Components; also makes eries of announcements), TPC -4.9%, WULF -4.4%, WULF -4.4%, PACS -4%, NNI -3.3%, MP -3.1%, DV -2.9% (also authorizes new $300 mln share repurchase program), TLN -2.8%, TLN -2.8%, UUUU -2.1%, ACA -2%, MNST -1.9%, RUN -1.5%, DFTX -1.4%, SNDX -1.3%, COMP -1% (also strategic alliance with RKT), OS -0.8%, AES -0.6%, CPNG -0.6%, WPM -0.6%, SOLV -0.5%, DNLI -0.5%, SOUN -0.3%

Companies trading lower in after hours in reaction to news: DUOT -12.2% (stock offering), KTOS -4.4% ($1 bln stock offering; also files for mixed shelf offering), BRSP -2.2% (files mixed securities shelf offering), PCOR -1.5% (PCOR and URI announce a new strategic partnership), WBD -1.3% (WBD determines revised proposal from PSKY constitutes a superior roposal; NFLX declines to match offer), META -0.7% (Google strikes AI chip deal with Meta, which sharpens Nvidia rivalry, according to TheInfo; also internal chip design efforts run into obstacles, according to The Information), MTCH -0.6% (files mixed securities shelf offering), JHG -0.6% (confirms receipt of unsolicited, non-binding proposal), EA -0.5% (did not secure bondholder support for buyback before deadline, according to Bloomberg), RYTM -0.4% (files mixed securities shelf offering; also files for offering by the selling shareholders), REXR -0.2% (names new COO), GL -0.1% (increases dividend)

TechCrunch : Everything announced at Samsung’s Galaxy Unpacked event, including

Everything announced at Samsung’s Galaxy Unpacked event, including S26 smartphones, Privacy Screen and more

Samsung’s Galaxy Unpacked event in San Francisco on Wednesday saw the Korean tech giant unveil its new S26 series of Galaxy smartphones, a pair of new earbuds, and talk a lot about AI assistants and agents.

Here’s a quick recap of everything announced at the event.

Galaxy S26 smartphones
The South Korean tech giant has lined up three new smartphones for the year: Galaxy S26 Ultra, Galaxy S26 Plus, and Galaxy S26, all of which feature iterative updates compared to their predecessors.

As the flagship, the S26 Ultra comes with Qualcomm’s latest Snapdragon 8 Elite Gen 5 processor, a 6.9-inch QHD+ display, and a 5,000 mAh battery. Samsung says the phone can charge from 0% to 75% in 30 minutes with a 60W charger, and supports the S-Pen stylus.
Image credit: Samsung
Samsung said while the S26 Ultra’s wide (200-megapixel) and telephoto (50-megapixel) lenses retain the same pixel count as its predecessor, the new wide camera has a bigger aperture of f/1.4, as does the telephoto lens, with an aperture of f/2.9.

The other two phones have the same processor as the S26 Ultra, but Samsung is using its own Exynos 2600 processor for these devices in some geographies. The smaller Galaxy S26 gets a battery bump from last year’s phone, and the S26 Plus can charge wirelessly at 20W.

The Galaxy S26 starts at $899 ($100 more than last year’s model), the Galaxy S26 Plus starts at $1,099 (also $100 more from last year), and the Galaxy S26 Ultra starts at $1,200.

Privacy display is the marquee feature for Ultra
Besides the hardware upgrades, Samsung has made a neat update to the S26 Ultra’s display that prevents people around you from reading what’s on the phone’s screen.
Users can choose to hide certain parts of the screen, like the notification area or the password field in login forms, or all of it. The privacy display can be configured for each app.
Image Credit: SamsungImage Credits:Samsung
Samsung said there is a “maximum privacy protection” feature as well, which tones down bright areas and lifts up dark parts of the screen.

Gemini’s agentic makeover and new ‘Circle to Search’
Google also made two significant announcements at Samsung’s event. First, the search giant previewed a new agentic version of Gemini that can perform tasks on your behalf, such as hailing a cab from Uber.

Samir Samat, head of Android ecosystem at Google, showed a demo where Gemini looks at a group chat and figures out the meal order that it should place on Grubhub.

The company also introduced a new version of “Circle to Search” that uses multi-object recognition to search for anything that you highlight on your phone’s display.

AI features
Earlier this week, Samsung said it was partnering with Perplexity to pre-load the AI company’s app on its smartphones and use its APIs to do stuff like set alarms or take notes, as well as power its browser’s search capabilities.

Samsung’s smartphones now come with three AI assistants: Bixby, Google Gemini, and Perplexity. The company is also using Galaxy AI to screen calls and give you a summary of what the caller says.

Samsung Galaxy Buds4
Samsung has also released two new earphones: the Galaxy Buds4 and the Galaxy Buds4 Pro. These earphones have a flatter stem design than their predecessors, and are rated IP54 and IP57, which means they are somewhat protected from dust and water.

Samsung says the Buds4 Pro has a new 11mm woofer, which widens the speaker area by 20%. The Pro models also feature a slightly longer battery life without the case compared to the regular version.
The Buds4 are priced at $179 while the Buds4 Pro cost $250.
Topics
g

The Information : A $1 Billion Payday From Google For Battery Startup Form Energ

A $1 Billion Payday From Google For Battery Startup Form Energy

Since it launched nine years ago, next-generation battery developer Form Energy has faced a single skeptical question: Do businesses really need a battery that lasts 100 hours or longer?

Google appeared to answer that question this week with a deal that will pay Form roughly $1 billion for a sprawling, 30 gigawatt-hour battery system to back up a new data center to be built later this decade in Minnesota. That is enough capacity to power 210,000 homes for four days.

On Tuesday, Google announced an agreement with power utility Xcel Energy to provide 1.9 GW of wind, solar and battery power for the data center, which the tech giant will build in Pine Island, Minn. Xcel will use Form’s iron-air batteries for long-term backup and standard lithium-ion batteries for instant surges of power.

Google’s move comes as President Donald Trump has pressured big tech companies to supply the power for the AI data centers they are building. Trump repeated the demand in his State of the Union address Tuesday.

Google and Xcel did not disclose the deal’s total dollar value, but Form CEO Mateo Jaramillo said the cost of his company’s piece of the project was roughly $1 billion.

The deal could mark the bottom of a dire period for next-generation Western battery startups. For more than two years, they have generally had difficulty raising new funds, leaving many struggling to survive and some going bankrupt. It’s a long way from the wave of initial public offerings these firms saw in 2020 and 2021.

Form is currently attempting to raise a $500 million round, Jaramillo said, and he plans to take the startup public next year. He declined to disclose what valuation he is raising the funds at.

Form has raised $1.4 billion to date. It last raised money at a $3.4 billion valuation in late 2024. Its lead investors for that round were Breakthrough Energy, Capricorn Investment Group and GE Vernova.

It was the first time Jaramillo confirmed the size of his fundraise, along with the timeline for Form’s IPO. Axios previously reported Form’s fundraise and its hopes to go public, citing anonymous sources.

Jaramillo, who previously worked at Tesla, where he helped to launch the Powerwall, its first stationary storage battery, said the Google deal will help him with the current fundraise since the database project is related to AI.

The deal squashes at least some skepticism about Form. The doubts have stemmed from a couple of things. For one, there are no current large-scale commercial deployments of iron-air batteries, the type Form has developed and is ramping up. In addition, many industry veterans have questioned whether there will be much demand for large batteries whose main selling point is that they last for four days.

“It really took this project to have [our thesis] externally validated,” Jaramillo said.

What sets an iron-air battery apart is both the cheapness of iron, one of the most plentiful metals on the planet, and how it works. Essentially, the battery derives energy by allowing iron to turn to rust, and then reversing it back into iron.

The process is massively inefficient. The battery loses up to half its energy in the charge-discharge cycle—a standard lithium-ion battery loses just a few percentage points. But Form has argued that an iron-air battery is so cheap and lasts so long that the energy loss is justified. Matched with renewable energy, which tends to deliver excess power at certain times, the lost energy hardly matters.

Most stationary storage batteries last around four hours. Jaramillo argues that at 100 hours of duration, Form’s batteries ought to earn consideration as a form of reliable base power, as natural gas and nuclear power plants are.

As an example, he said that an iron-air battery would come to good use when extended extreme weather conditions stretch grid capacity to its maximum. Over a 10-day period, the utility could draw power from the battery for 10 hours a day. The battery could keep delivering power without a recharge, “because you don’t want to add incremental stress to the grid,” he said.

Another part of the justification for Form’s batteries is that the federal government is subsidizing it. In 2024, the Department of Energy awarded the company a $150 million grant to expand its Weirton, W.Va., factory to an annual production capacity of 500 MW. Jaramillo said Form would begin production at the plant in 2028, and working at full capacity it will be able to fulfill the Google contract in about seven months.

In addition, Jaramillo said, the Google project qualifies for a significant tax credit under the Inflation Reduction Act, passed under former President Joe Biden. The base credit would be 30%, but the credits could rise to half the project’s costs because it’s using U.S.-made iron and will be located on the site of a coal plant.

The Information : Alphabet and Other Big Tech Could Borrow Hundreds of Billions

Alphabet and Other Big Tech Could Borrow Hundreds of Billions Each

The Takeaway
  • Big Tech firms could borrow up to $200 billion each while maintaining credit ratings.
  • AI investments are transforming Big Tech into more indebted companies.
  • Credit ratings for tech giants appear safe for two years, but risks exist.

Amazon, Alphabet and Meta Platforms in recent months have each turned to the bond markets to raise tens of billions of dollars for their massive investments in AI data centers. And all three are likely to borrow much more in the next couple of years, with their projected capital expenditures now likely to come close to or surpass the cash they generate.

How much more can they borrow, without suffering a credit ratings downgrade and seeing their financing costs rise, is now a big question. Right now, credit agency S&P estimates all three will end up with a little more debt than cash by the end of this year, the reverse of the current situation. Longer term, however, the companies could each borrow close to $200 billion and still retain their credit rating, judging by S&P’s methodology.

Companies’ newfound emphasis on borrowings to fund expansion show how the AI boom will likely forever change the financial profile of the biggest tech companies. Instead of generating vast amounts of cash that allow for diversification into other businesses and giant stock buybacks, tech firms appear likely to be transformed into heavily indebted companies producing little free cash—at least for the next few years.

S&P and Moody’s calculate credit ratings by looking at the companies’ current debt and earnings alongside forecasts of how that ratio might change over time.

Right now, both agencies give Alphabet a higher investment grade rating than either Meta or Amazon: an AA+ or AA2. Meta has the lowest—AA- by S&P and AA3 by Moody’s. That reflects a belief among credit analysts that Meta’s reliance on advertising means it isn’t as diversified as Alphabet and Amazon—both of which have bountiful cloud businesses, among other lucrative parts of their operations.

Both credit agencies have indicated they don’t expect the tech companies’ credit ratings to change in the next couple of years. Jawad Hussain, director at S&P Global Ratings, says Alphabet’s AA+ rating, for instance, is probably safe until the company passes a threshold of one times debt-to-earnings before interest, taxes, depreciation and amortization. Right now, Alphabet has a ratio of zero, because it has more cash than debt.

Given that analysts estimate Alphabet will generate $216 billion in Ebitda this year, a one times debt to Ebitda ratio implies Alphabet would have roughly $200 billion of debt, after deducting cash on its balance sheet. At the end of 2025, the company had $46 billion debt and $126 billion in cash.

The company has projected it will spend about $180 billion in capital expenditures this year, however, absorbing most of the cash it is expected to generate. But Alphabet also is scheduled to complete its acquisitions of Wiz, a cybersecurity firm, and Intersect, a data center infrastructure company, costing it a combined $37 billion.

As a result, S&P predicts that Alphabet will end 2026 with $16 billion more debt than cash, translating to a debt to ebitda ratio of between 0.1 and 0.2 times. That assumes Alphabet finishes the year with total debt of $117 billion, including future data center leases, which S&P includes in its debt calculation. It assumes the company will finish the year with $102 billion in cash.

The agency’s latest forecasts don’t show any of the three companies—Alphabet, Amazon or Meta—exceeding a ratio of 0.5 times debt to ebitda in either 2026 or 2027. That implies its credit ratings are safe for a couple of years.

However, one risk for the companies is that S&P “might rethink” the downgrade threshold of one times debt-to-EBITDA, according to Hussein. It might do that if lower cash flows become commonplace or if S&P decides the companies are not reaping adequate returns on their AI investments.

Christian Hoffmann, head of fixed income at Thornburg Investment Management, said a downgrade for any of these companies is “certainly a possibility” given their planned increases in capital spending this year, though he added that “the cost of debt is not high and frankly, even if they carried a lower credit rating, it would still not be terribly high.”

Hoffmann acknowledged that while demand for tech company debt is “pretty high” right now, “at some point, I think people decide that they have had enough.” That would manifest itself in tech debt spreads—the gap between the interest costs on the bonds and Treasurys—widening.

One signal that this might be happening already is that the margin over the cost of Treasury notes on Meta’s credit default swaps, financial instruments akin to insurance on a company’s debt, has risen significantly over the last year.