Connect with us

Business

Bloom Energy’s stock is up 1,000% in a year because its fuel cells are solving AI’s data center power problem

Published

on


Aerospace engineer KR Sridhar always dreamed big: He used to work with NASA on technology to convert carbon dioxide into oxygen to support life on other planets or let humans breathe air on Mars. But as the Soviet Union fell and the space race slowed, Sridhar pivoted to providing clean energy technology for the rising global middle class.

He cofounded Ion America in 2001—renamed Bloom Energy five years later—with a focus on fuel cells that deliver cleaner, on-site, off-grid power. Fast forward to today’s AI race, and Bloom’s products just so happen to mesh with the needs of the data center boom that’s starving for massive power generation growth very quickly.

Fuel cells can hypothetically bring power online for data centers in months, not years, because they do not have to wait for the backlog of gas-fired turbines or the long queue for grid interconnections.

Bloom’s stock price has spiked 1,000% in 12 months—its market cap is now about $28 billion, up from $2.5 billion a year ago. The company has signed big data center deals with Oracle, American Electric Power (AEP), Equinix, and Brookfield Asset Management, the latter of which is a $5 billion partnership announced Oct. 13 to power AI factories globally, including Europe.

Bloom CEO Sridhar actually had data centers in mind as a big opportunity when he first pitched the company at the turn of the century. But the massive growth didn’t take off until after the ChatGPT launch.

“That’s when we said, ‘Everything that we’ve been telling the world is going to happen is now going to accelerate,’” Sridhar told Fortune.

“It’s a 24-year journey for an overnight success,” Sridhar said with a laugh. “I’m glad it’s in my lifetime.”

Renewable wind and solar energy still have some intermittency issues, even with batteries. And a sufficient supply of gas-fired and nuclear power stations are several years away, he said. That’s why on-site fuel cells are the answer, he says, both as a bridge and as a permanent power solution.

The company’s solid oxide fuel cells are a mature technology that have been developed over two decades. Thus far, Bloom has deployed 1.5 gigawatts of fuel cells—enough to power 1.2 million homes—with demand mounting by the day. The goal is to deploy 10 gigawatts per year from its manufacturing hubs in Fremont, California, and Newark, Delaware.

Fuel cells have existed for years, but they’ve lacked mainstream adoption because of their high manufacturing costs. They require expensive precious metals, corrosive acids, or hard-to-contain molten materials. Bloom’s solid oxide fuel cells use lower-cost ceramics—no precious metals—and they provide much greater electrical efficiency, operating at temperatures above 800 degrees Celsius.

The cells convert natural gas, hydrogen, or biogas into electricity through a clean electro-chemical process rather than dirty combustion. The cells are zero-carbon if they use green hydrogen, but they’re still cleaner than gas turbines even if they use natural gas. And the fuel cells are modular, so they can ramp up or down, or be relocated to other data centers when grid power becomes available.

Sridhar acknowledged the long journey to get here. Bloom took seven years to develop the first commercial cells. And then another decade to continually bring the costs down and improve their efficiency. In the meantime, Bloom relied on “early adopter” Fortune 100 customers who were willing to pay extra to power for cleaner power, including Google, Walmart, eBay, and FedEx.

From unicorn to large-cap stock

For years, Bloom was hyped as a Silicon Valley unicorn, but in 2012 the SEC charged an investment bank working with Bloom of using inflated numbers to mislead investors. Bloom was not accused of wrongdoing, and the company eventually went public in 2018.

The technology works more affordably now since fuel cell microgrids qualify for tax credits from President Trump’s One Big Beautiful Bill, said Marina Domingues, head of U.S. new energies research for the Rystad Energy research firm. She said they are comparable to the price of power from combined-cycle gas turbines, but fuel cells can come online more quickly and produce power more cleanly.

“Deta centers come with two main wish list requests. One of them is the power must be truly reliable,” she said. “Another one, which is probably the toughest, is that they need power now,” Domingues said. “They’re offering a solution exactly at the same time developers need it. There’s a lot of potential market growth for a company like Bloom.”

Bloom Energy explains how its fuel cells function.

Pushing a fuel cell imperative

Much of the AI boom’s focus is on massive hyperscaler campuses in rural areas, such as OpenAI’s Stargate project in Abilene, Texas, and beyond. But Sridhar insists fuel cells will not only be helpful, they’ll become imperative once the race is on to build more and more smaller data centers in increasingly urban areas closer to consumer demand.

“The only two raw materials [that AI needs] are data and electricity,” Sridhar said. “It’s extremely electricity intense. They have to produce their own power on site.

“You’re not going to have any choice but on-site power, because no city has the distribution network that can accommodate those kinds of big [electricity] loads,” Sridhar said, citing Memphis, Tennessee, as an example of public outcry amid rising emissions for powering data centers. “If it’s in your backyard or outside your office window, you want it to be clean.”

Elham Akhavan, Wood Mackenzie senior microgrid research analyst, said Bloom’s competitors—including FuelCell Energy, Doosan Group’s HyAxiom, and Plug Power—offer different variations on the technology but they have not yet scaled up as much as Bloom.

As the technology advanced, the fuel cell sector was able to reposition itself from being a mere provider of backup power to a primary power source—with the grid as the backup, she said.

“Bloom led fuel cell deployment across North America way before data center demand arrived,” Akhavan said. “It’s a prime power solution in a very small footprint, and rest of the land is available for the data centers.”

Domingues said Bloom has a multitude of factors working to its advantage, including a head start on competitors, a domestic manufacturing chain when Trump is pushing onshoring and tariffs, and an early bet on fuel cells for “stationary power” when many potential rivals focused on fuel cells for the transportation sector.

“Bloom bet on the stationary power path, and they also had strong relationships with some of the traditional data center market,” Domingues said. “That allows them some competitive advantages against their peers.”

The company is currently unprofitable. Bloom operated at a $29 million net loss in 2024, improved from a roughly $300 million loss in 2023. But Bloom also lost $66 million in the first half of 2025.

Sridhar insists the lack of profitability will be short lived. “We are not one of these companies that has to invest, invest, invest. We’ve already done that part the last 20 years. We have the flywheel spinning already. Accelerating is going to take less and less energy.”

Sridhar said Bloom intentionally built its manufacturing plants as exact copies so they can continue to scale up more quickly to match demand and offer rapid returns on investment. “We believe the market demand is going to be there because electricity abundance is what’s going to generate a better quality of life and wealth in a digitized world,” he said.

Sridhar is still inspired by his dreams of Mars: “Living off the land is truly what an explorer does. So, I started producing oxygen, breathing air, water, electricity, heat on Mars so someday humans can live there,” he said. On Earth, “Clean energy is what we need on the planet in a very reliable way, everywhere, and with access.”



Source link

Continue Reading

Business

YouTube launches option for U.S. creators to receive stablecoin payouts through PayPal

Published

on



Big Tech continues to tiptoe into crypto. The latest example is a move by YouTube to let creators on the video platform choose to receive payouts in PayPal’s stablecoin. The head of crypto at PayPal, May Zabaneh, confirmed the arrangement to Fortune, adding that the feature is live and, as of now, only applies to users in the U.S. 

A spokesperson for Google, which owns YouTube, confirmed the video site has added payouts for creators in PayPal’s stablecoin but declined to comment further.

YouTube is already an existing customer of PayPal’s and uses the fintech giant’s payouts service, which helps large enterprises pay gig workers and contractors. 

Early in the third quarter, PayPal added the capability for payment recipients to receive their checks in PayPal’s stablecoin, PYUSD. Afterwards, YouTube decided to give that option to creators, who receive a share of earnings from the content they post on the platform, said Zabaneh.

“The beauty of what we’ve built is that YouTube doesn’t have to touch crypto and so we can help take away that complexity,” she added.

Big Tech eyes stablecoins

YouTube’s interest in stablecoins comes as Google and other Big Tech companies have shown interest in the cryptocurrencies amid a wave of hype in Silicon Valley and beyond. 

The tokens, which are pegged to underlying assets like the U.S. dollar, are longtime features of the crypto industry. But over the past year, they’ve exploded into the mainstream, especially after President Donald Trump signed into law a new bill regulating the crypto assets. Proponents say they are an upgrade over existing financial infrastructure, and big fintechs have taken notice, including Stripe. In February, the payments giant closed a blockbuster $1.1 billion purchase of the stablecoin startup Bridge.

PayPal has long been an earlier mover in crypto among large tech firms. In 2020, it let users buy and sell Bitcoin, Ethereum, and a handful of other cryptocurrencies. And, in 2023, it launched the PYSUD stablecoin, which now has a market capitalization of nearly $4 billion, according to CoinGecko.

PayPal has slowly integrated PYUSD throughout its stable of products. Users can hold it in its digital wallet as well as Venmo, another financial app that PayPal also owns. They can use it to pay merchants. And, in February, a PayPal executive said small-to-medium sized merchants will be able to use it to pay vendors.

YouTube’s addition of payouts in PYUSD isn’t the first time Google has experimented with PayPal’s stablecoin. An executive at Google Cloud, the tech giant’s cloud computing arm, previously toldFortune that it had received payments from two of its customers in PYUSD. 



Source link

Continue Reading

Business

Oracle slides by most since January on mounting AI spending

Published

on



Oracle Corp. shares plunged the most in almost 11 months after the company escalated its spending on AI data centers and other equipment, rising outlays that are taking longer to translate into cloud revenue than investors want.

Capital expenditures, a metric of data center spending, were about $12 billion in the quarter, an increase from $8.5 billion in the preceding period, the company said Wednesday in a statement. Analysts anticipated $8.25 billion in capital spending in the quarter, according to data compiled by Bloomberg. 

Oracle now expects capital expenditures will reach about $50 billion in the fiscal year ending in May 2026 — a $15 billion increase from its September forecast — executives said on a conference call after the results were released.

The shares fell 11% to $198.85 at the close Thursday in New York, the biggest single-day decline since Jan. 27. Oracle’s stock had already lost about a third of its value through Wednesday’s close since a record high on Sept. 10. Meanwhile, a measure of Oracle’s credit risk reached a fresh 16-year high.

The latest earning report and share slide marks a reversal of fortunes for a company that just a few months ago was enjoying a blistering rally and clinching multibillion-dollar data center deals with the likes of OpenAI. The gains temporarily turned co-founder Larry Ellison into the world’s richest person, with the tech magnate passing Elon Musk for a few hours.

Known for its database software, Oracle has recently found success in the competitive cloud computing market. It’s engaging in a massive data center build-out to power AI work for OpenAI and also counts companies such as ByteDance Ltd.’s TikTok and Meta Platforms Inc. as major cloud customers. 

Fiscal second-quarter cloud sales increased 34% to $7.98 billion, while revenue in the company’s closely watched infrastructure business gained 68% to $4.08 billion. Both numbers fell just short of analysts’ estimates.Play Video

Still, Wall Street has raised doubts about the costs and time required to develop AI infrastructure at such a massive scale. Oracle has taken out significant sums of debt and committed to leasing multiple data center sites. 

The cost of protecting the company’s debt against default for five years rose as much as 0.17 percentage point to around 1.41 percentage point a year, the highest intraday level since April 2009, according to ICE Data Services. The gauge rises as investor confidence in the company’s credit quality falls. Oracle credit derivatives have become a credit market barometer for AI risk.

“Oracle faces its own mounting scrutiny over a debt-fueled data center build-out and concentration risk amid questions over the outcome of AI spending uncertainty,” said Jacob Bourne, an analyst at Emarketer. “This revenue miss will likely exacerbate concerns among already cautious investors about its OpenAI deal and its aggressive AI spending.”

Remaining performance obligation, a measure of bookings, jumped more than fivefold to $523 billion in the quarter, which ended Nov. 30. Analysts, on average, estimated $519 billion.

Investors want to see Oracle turn its higher spending on infrastructure into revenue as quickly as it has promised. 

“The vast majority of our cap ex investments are for revenue generating equipment that is going into our data centers and not for land, buildings or power that collectively are covered via leases,” Principal Financial Officer Doug Kehring said on the call. “Oracle does not pay for these leases until the completed data centers and accompanying utilities are delivered to us.”

“As a foundational principle, we expect and are committed to maintaining our investment grade debt rating,” Kehring added.

Oracle’s cash burn increased in the quarter and its free cash flow reached a negative $10 billion. Overall, the company has about $106 billion in debt, according to data compiled by Bloomberg. “Investors continually seem to expect incremental cap ex to drive incremental revenue faster than the current reality,” wrote Mark Murphy, an analyst at JP Morgan.Play Video

“Oracle is very good at building and running high-performance and cost-efficient cloud data centers,” Clay Magouyrk, one of Oracle’s two chief executive officers, said in the statement. “Because our data centers are highly automated, we can build and run more of them.”

This is Oracle’s first earnings report since longtime Chief Executive Officer Safra Catz was succeeded by Magouyrk and Mike Sicilia, who are sharing the CEO post.

Part of the negative sentiment from investors in recent weeks is tied to increased skepticism about the business prospects of OpenAI, which is seeing more competition from companies like Alphabet Inc.’s Google, wrote Kirk Materne, an analyst at Evercore ISI, in a note ahead of earnings. Investors would like to see Oracle management explain how they could adjust spending plans if demand from OpenAI changes, he added.

In the quarter, total revenue expanded 14% to $16.1 billion. The company’s cloud software application business rose 11% to $3.9 billion. This is the first quarter that Oracle’s cloud infrastructure unit generated more sales than the applications business.

Earnings, excluding some items, were $2.26 a share. The profit was helped by the sale of Oracle’s holdings in chipmaker Ampere Computing, the company said. That generated a pretax gain of $2.7 billion in the period. Ampere, which was backed early in its life by Oracle, was bought by Japan’s SoftBank Group Corp. in a transaction that closed last month.

In the current period, which ends in February, total revenue will increase 19% to 22%, while cloud sales will increase 40% to 44%, Kehring said on the call. Both forecasts were in line with analysts’ estimates.

Annual revenue will be $67 billion, affirming an outlook the company gave in October.



Source link

Continue Reading

Business

Analyst sees Disney/OpenAI deal as a dividing line in entertainment history

Published

on



Disney’s expansive $1 billion licensing agreement with OpenAI is a sign Hollywood is serious about adapting entertainment to the age of artificial intelligence (AI), marking the start of what one Ark Invest analyst describes as a “pre‑ and post‑AI” era for entertainment content. The deal, which allows OpenAI’s Sora video model to use Disney characters and franchises, instantly turns a century of carefully guarded intellectual property (IP) into raw material for a new kind of crowd‑sourced, AI‑assisted creativity.​

Nicholas Grous, director of research for consumer internet and fintech at Ark Invest, told Fortune tools like Sora effectively recreate the “YouTube moment” for video production, handing professional‑grade creation capabilities to anyone with a prompt instead of a studio budget. In his view, that shift will flood the market with AI‑generated clips and series, making it far harder for any single new creator or franchise to break out than it was in the early social‑video era.​ His remarks echoed the analysis from Melissa Otto, head of research at S&P Global Visible Alpha, who recently told Fortune Netflix’s big move for Warner Bros.’ reveals the streaming giant is motivated by a need to deepen its war chest as it sees Google’s AI-video capabilities exploding with the onset of TPU chips.

As low‑cost synthetic video proliferates, Grous said he believes audiences will begin to mentally divide entertainment into “pre‑AI” and “post‑AI” categories, attaching a premium to work made largely by humans before generative tools became ubiquitous. “I think you’re going to have basically a split between pre-AI content and post-AI content,” adding that viewers will consider pre-AI content closer to “true art, that was made with just human ingenuity and creativity, not this AI slop, for lack of a better word.”

Disney’s IP as AI fuel

Within that framework, Grous argued Disney’s real advantage is not just Sora access, but the depth of its pre‑AI catalog across animation, live‑action films, and television. Iconic franchises like Star Wars, classic princess films and legacy animated characters become building blocks for a global experiment in AI‑assisted storytelling, with fans effectively test‑marketing new scenarios at scale.​

“I actually think, and this might be counterintuitive, that the pre-AI content that existed, the Harry Potter, the Star Wars, all of the content that we’ve grown up with … that actually becomes incrementally more valuable to the entertainment landscape,” Grous said. On the one hand, he said, there are deals like Disney and OpenAI’s where IP can become user-generated content, but on the other, IP represents a robust content pipeline for future shows, movies, and the like.

Grous sketched a feedback loop in which Disney can watch what AI‑generated character combinations or story setups resonate online, then selectively “pull up” the most promising concepts into professionally produced, higher‑budget projects for Disney+ or theatrical release. From Disney’s perspective, he added, “we didn’t know Cinderella walking down Broadway and interacting with these types of characters, whatever it may be, was something that our audience would be interested in.” The OpenAI deal is exciting because Disney can bring that content onto its streaming arm Disney+ and make it more premium. “We’re going to use our studio chops to build this into something that’s a bit more luxury than what just an individual can create.”

Grous agreed the emerging market for pre‑AI film and TV libraries is similar to what’s happened in the music business, where legacy catalogs from artists like Bruce Springsteen and Bob Dylan have fetched huge sums from buyers betting on long‑term streaming and licensing value.

The big Netflix-Warner deal

For streaming rivals, the Disney-OpenAI pact is a strategic warning shot. Grous argued the soaring price tags in the bidding war for Warner Bros. between Netflix and Paramount shows the importance of IP for the next phase of entertainment. “​I think the reason this bidding [for Warner Bros.] is approaching $100 billion-plus is the content library and the potential to do a Disney-OpenAI type of deal.” In other words, whoever controls Batman and the like will control the inevitable AI-generated versions of those characters, although “they could take a franchise like Harry Potter and then just create slop around it.”

Netflix has a great track record on monetizing libraries, Grous said, listing the example of how the defunct USA dramedy Suits surged in popularity once it landed on Netflix, proving extensive back catalogs can be revived and re‑monetized when matched with modern distribution.​

Grous cited Nintendo and Pokémon as examples of under‑monetized franchises that could see similar upside if their owners strike Sora‑style deals to bring characters more deeply into mobile and social environments.​ “That’s another company where you go, ‘Oh my god, the franchises they have, if they’re able to bring it into this new age that we’re all experiencing, this is a home-run opportunity.’”

In that environment, the Ark analyst suggests Disney’s OpenAI deal is less of a one‑off licensing win than an early template for how legacy media owners might survive and thrive in an AI‑saturated market. The companies with rich pre‑AI catalogs and a willingness to experiment with new tools, he argued, will be best positioned to stand out amid the “AI slop” and turn nostalgia‑laden IP into enduring, flexible assets for the post‑AI age.​

Underlying all of this is a broader battle for attention that spans far beyond traditional studios and shows how sectors between tech and entertainment are getting even blurrier than when the gatecrashers from Silicon Valley first piled into streaming. Grous notes Netflix itself has long framed its competition as everything from TikTok and Instagram to Fortnite and “sleep,” a mindset that fits naturally with the coming wave of AI‑generated video and interactive experiences.​ (In 2017, Netflix co-founder Reed Hastings famously said “sleep” was one of the company’s biggest competitors, as it was busy pioneering the binge-watch.)

Grous also sounded a warning for the age of post-AI content: The binge-watch won’t feel as good anymore, and there will be some kind of backlash. As critics such as The New York Times‘ James Poniewozik increasingly note, streaming shows don’t seem to be as re-watchable as even recent hits from the golden age of cable TV, such as Mad Men. Grous said he sees a future where the endangered movie theater makes a comeback. “People are going to want to go outside and meet or go to the theater. Like, we’re not just going to want to be fed AI slop for 16 hours a day.”

Editor’s note: the author worked for Netflix from June 2024 through July 2025.



Source link

Continue Reading

Trending

Copyright © Miami Select.