Connect with us

Business

Companies like OpenAI are sucking up power at a historic rate. One startup thinks it has found a way to take pressure off the grid

Published

2 months ago

on



The numbers are nothing short of staggering. Take Sam Altman, Open AI’s CEO. He reportedly wants 250 gigawatts of new electricity—equal to about half of Europe’s all-time peak load—to run gigantic new data centers in the U.S. and elsewhere worldwide by 2033.

Building or expanding power plants to generate that much electricity on Altman’s timetable indeed seems almost inconceivable. “What OpenAI is trying to do is absolutely historic,” says Virun Sivaram, Senior Fellow for Energy and Climate at the Council on Foreign Relations. The problem is, “there is no way today that our grids, with our power plants, can supply that energy to those projects, and it can’t possibly happen on the timescale that AI is trying to accomplish.”

Yet Sivaram believes Altman may be able to reach his goal of running multiple new data centers in a different way. Sivaram, in addition to his position at the CFR, is the founder and CEO of Emerald AI, a startup that launched in July. “I founded it directly to solve this problem,” he says—not just Altman’s problem specifically, but the larger problem of powering the data centers that all AI companies need. Several smart minds in tech like the odds of Sivaram’s company. It’s backed by Radical Ventures, Nvidia’s venture capital arm NVentures, other VCs, and heavy-hitter individuals including Google chief scientist Jeff Dean and Kleiner Perkins chairman John Doerr.

Emerald AI’s premise is that the electricity needed for AI data centers is largely there already. Even big new data centers would confront power shortages only occasionally. “The power grid is kind of like a superhighway that faces peak rush hour just a few hours per month,” Sivaram says. Similarly, in most places today the existing grid could handle a data center easily except in a few times of extreme demand.

Sivaram’s objective is to solve the problem of those rare high-demand moments the grid can’t handle. It isn’t all that difficult, at least in theory, he argues. Some jobs can be paused or slowed, he explains, like the training or fine-tuning of a large language model for academic research. Other jobs, like queries for an AI service used by millions of people, can’t be rescheduled but could be redirected to another data center where the local power grid is less stressed. Data centers would need to be flexible in this way less than 2% of the time, he says; Emerald AI is intended to help them do it by turning the theory to real-world action. The result, Sivaram says, would be profound: “If all AI data centers ran this way, we could achieve Sam Altman’s global goal today.”

A paper by Duke University scholars, published in February, reported a test of the concept and found it worked. Separately, Emerald AI and Oracle tried the concept on a hot day in Phoenix and found they could reduce power consumption in a way that didn’t degrade AI computation—“kind of having your cake and eating it too,” Sivaram says. That paper is under peer review.

No one knows if Altman’s 250-gigawatt plan will prove to be brilliant or folly. In these early days, Emerald AI’s future can’t be divined, as promising as it seems. What we know for sure is that great challenges bring forth unimagined innovations—and in the AI era, we should brace for plenty of them.

Fortune Global Forum returns Oct. 26–27, 2025 in Riyadh. CEOs and global leaders will gather for a dynamic, invitation-only event shaping the future of business. Apply for an invitation.



Source link

Related Topics:
Continue Reading

Business

Databricks CEO Ali Ghodsi says company will be worth $1 trillion by doing these three things

Published

3 hours ago

on

December 9, 2025

By



Ali Ghodsi, the CEO and cofounder of data intelligence company Databricks, is betting his privately held startup can be the latest addition to the trillion-dollar valuation club.

In August, Ghodsi told the Wall Street Journalthat he believed Databricks, which is reportedly in talks toraise funding at a $134 billion valuation, had “a shot to be a trillion-dollar company.” At Fortune’s Brainstorm AI conference in San Francisco on Tuesday, he explained how it would happen, laying out a “trifecta” of growth areas to ignite the company’s next leg of growth.

The first is entering the transactional database market, the traditional territory of large enterprise players like Oracle, which Ghodsi said has remained largely “the same for 40 years.” Earlier this year, Databricks launched a link-based offering called Lakehouse, which aims to combine the capabilities of traditional databases with modern data lake storage, in an attempt to capture some of this market.

The company is also seeing growth driven by the rise of AI-powered coding. “Over 80% of the databases that are being launched on Databricks are not being launched by humans, but by AI agents,” Ghodsi said. As developers use AI tools for “vibe coding”—rapidly building software with natural language commands—those applications automatically need databases, and Ghodsi they’re defaulting to Databricks’ platform.

“That’s just a huge growth factor for us. I think if we just did that, we could maybe get all the way to a trillion,” he said.

The second growth area is Agentbricks, Databricks’ platform for building AI agents that work with proprietary enterprise data.

“It’s a commodity now to have AI that has general knowledge,” Ghodsi said, but “it’s very elusive to get AI that really works and understands that proprietary data that’s inside enterprise.” He pointed to the Royal Bank of Canada, which built AI agents for equity research analysts, as an example. Ghodsi said these agents were able to automatically gather earnings calls and company information to assemble research reports, reducing “many days’ worth of work down to minutes.”

And finally, the third piece to Ghodsi’s puzzle involves building applications on top of this infrastructure, with developers using AI tools to quickly build applications that run on Lakehouse and which are then powered by AI agents. “To get the trifecta is also to have apps on top of this. Now you have apps that are vibe coded with the database, Lakehouse, and with agents,” Ghodsi said. “Those are three new vectors for us.”

Ghodsi did not provide a timeframe for attaining the trillion-dollar goal. Currently, only a handful of companies have achieved the milestone, all of them as publicly traded companies. In the tech industry, only big tech giants like Apple, Microsoft, Nvidia, Alphabet, Amazon, and Meta have managed to cross the trillion-dollar threshold.

To reach this level would require Databricks, which is widely expected to go public sometime in early 2026, to grow its valuation roughly sevenfold from its current reported level. Part of this journey will likely also include the expected IPO, Ghodsi said.

“There are huge advantages and pros and cons. That’s why we’re not super religious about it,” Ghodsi said when asked about a potential IPO. “We will go public at some point. But to us, it’s not a really big deal.”

Could the company IPO next year? Maybe, replied Ghodsi.



Source link

Continue Reading

Business

New contract shows Palantir working on tech platform for another federal agency that works with ICE

Published

4 hours ago

on

December 9, 2025

By



Palantir, the artificial intelligence and data analytics company, has quietly started working on a tech platform for a federal immigration agency that has referred dozens of individuals to U.S. Immigration and Customs Enforcement for potential enforcement since September.

The U.S. Citizenship and Immigration Services agency—which handles services including citizenship applications, family immigration, adoptions, and work permits for non-citizens—started the contract with Palantir at the end of October, and is paying the data analytics company to implement “Phase 0” of a “vetting of wedding-based schemes,” or “VOWS” platform, according to the federal contract, which was posted to the U.S. government website and reviewed by Fortune.

The contract is small—less than $100,000—and details of what exactly the new platform entails are thin. The contract itself offers few details, apart from the general description of the platform (“vetting of wedding-based schemes”) and an estimate that the completion of the contract would be Dec. 9.Palantir declined to comment on the contract or nature of the work, and USCIS did not respond to requests for comment for this story.

But the contract is notable, nonetheless, as it marks the beginning of a new relationship between USCIS and Palantir, which has had longstanding contracts with ICE, another agency of the Department of Homeland Security, since at least 2011. The description of the contract suggests that the “VOWS” platform may very well be focused on marriage fraud and related to USCIS’ recent stated effort to drill down on duplicity in applications for marriage and family-based petitions, employment authorizations, and parole-related requests.

USCIS has been outspoken about its recent collaboration with ICE. Over nine days in September, USCIS announced that it worked with ICE and the Federal Bureau of Investigation to conduct what it called “Operation Twin Shield” in the Minneapolis-St. Paul area, where immigration officials investigated potential cases of fraud in immigration benefit applications the agency had received. The agency reported that its officers referred 42 cases to ICE over the period. In a statement published to the USCIS website shortly after the operation, USCIS director Joseph Edlow said his agency was “declaring an all-out war on immigration fraud” and that it would “relentlessly pursue everyone involved in undermining the integrity of our immigration system and laws.” 

“Under President Trump, we will leave no stone unturned,” he said.

Earlier this year, USCIS rolled out updates to its policy requirements for marriage-based green cards, which have included more details of relationship evidence and stricter interview requirements.

While Palantir has always been a controversial company—and one that tends to lean into that reputation no less—the new contract with USCIS is likely to lead to more public scrutiny. Backlash over Palantir’s contracts with ICE have intensified this year amid the Trump Administration’s crackdown on immigration and aggressive tactics used by ICE to detain immigrants that have gone viral on social media. Not to mention, Palantir inked a $30 million contract with ICE earlier this year to pilot a system that will track individuals who have elected to self-deport and help ICE with targeting and enforcement prioritization. There has been pushback from current and former employees of the company alike over contracts the company has with ICE and Israel.

In a recent interview at the New York Times DealBook Summit, Karp was asked on stage about Palantir’s work with ICE and later what Karp thought, from a moral standpoint, about families getting separated by ICE. “Of course I don’t like that, right? No one likes that. No American. This is the fairest, least bigoted, most open-minded culture in the world,” Karp said. But he said he cared about two issues politically: immigration and “re-establishing the deterrent capacity of America without being a colonialist neocon view. On those two issues, this president has performed.”



Source link

Continue Reading

Business

CoreWeave CEO: Despite see-sawing stock, IPO was ‘incredibly successful’ amid challenges of tariff timing

Published

4 hours ago

on

December 9, 2025

By



CoreWeave has been rocked by dizzying stock swings—with its stock currently trading 52% below its post-IPO high—and a frequent target of market commentators, but CEO Michael Intrator says the company’s move to the public markets has been “incredibly successful. And he takes the public’s mixed reaction in stride, given the novelty of CoreWeave’s “neocloud” business which competes with established cloud providers like Amazon AWS and Google Cloud.

“When you introduce new models, introduce a new way of doing business, disrupt what has been a static environment, it’s going to take some people some time,” Intrator said Tuesday at Fortune’s Brainstorm AI conference in San Francisco. But, he added, more people are beginning to understand the CoreWeave’s business model.

“We came out into one of the most challenging environments,” Intrator said of CoreWeave’s March IPO, which occurred very close to President Trump’s “Liberation Day” tariffs in April. “In spite of the incredible headwinds, we’re able to launch a successful IPO.”

CoreWeave, which priced its IPO at $40 per share, has experienced frequent severe up-and-down price swings in the eight months since its public market debut. At its closing price of $90.66 on Tuesday, the stock remains well above its IPO price.

As Fortune reported last month, CoreWeave’s rapid rise has been fueled by an aggressive, debt-heavy strategy to stand up data centers at unprecedented speed for AI customers. And for now, the bet is still paying off. In its third-quarter results released in November, the company said its revenue backlog nearly doubled in a single quarter—to $55.6 billion from $30 billion—reflecting long-term commitments from marquee clients including Meta, OpenAI, and French AI startup Poolside. Both earnings and revenue came in ahead of Wall Street expectations.

But the numbers were not all celebratory. CoreWeave disclosed a further increase in the debt it has taken on to finance its expansion, and it revised its full-year revenue outlook downward—suggesting that, even with historic demand in the pipeline.

With media headlines calling CoreWeave a “ticking time bomb,” with critics calling out insider stock sales, circular financing accusations and an overreliance on Nvidia, Intrator was asked whether he felt CoreWeave was misunderstood.

“Look, we built a company that is challenging one of the most stable businesses that exist—that cloud business, these three massive players,” he said, referring to AWS, Microsoft Azure and Google Cloud.  I feel like it’s incumbent on CoreWeave to introduce a new business model on how the cloud is going to be built and run. And that’s what we’re doing.” 

He repeatedly framed CoreWeave not as a GPU reseller or traditional data-center operator but as a company purpose-built from scratch to deliver high-performance, parallelized computing for AI workloads. That focus, he said, means designing proprietary software that orchestrates GPUs, building and colocating its own infrastructure, and moving “up the stack” through acquisitions such as Weights & Biases and OpenPipe.

Intrator also defended the company’s debt strategy, saying CoreWeave is effectively inventing a new financing model for AI infrastructure. He pointed to the company’s ability to repurpose power sources, rapidly deploy capacity, and finance large-scale clusters as proof it is solving problems incumbents never had to face.

“When I look back at history of the company, it took us a year with with a company investor like Fidelity, before they were like, ‘Oh, I get it,’” he said. “So look, we’ve been public for eight months. I couldn’t be prouder of what the company has accomplished.” 



Source link

Continue Reading

Trending

Copyright © Miami Select.