Elon Musk's xAI eyes $12 billion debt deal to train Grok on Nvidia chips, launch AI supercluster with 550,000 GPUs; details here
The financing deal is being spearheaded by Valor Equity Partners, an investment firm led by Musk's close associate Antonio Gracias. Talks are ongoing with multiple lenders to support the chip acquisition. Some of the lenders are seeking repayment within three years and want limits placed on the borrowed amount to reduce their financial risk.
Grok training on 230,000 GPUs and counting
The primary use of the funds will be to buy Nvidia chips, which will then be leased to xAI to power the infrastructure for Grok—xAI's AI chatbot. Musk posted on X that Grok is currently being trained on 230,000 graphics processing units (GPUs), including 30,000 of Nvidia's GB200 chips. Inference operations are being handled by cloud service providers.
Cable pr0n of @xAI GB200 servers at Colossus 2 pic.twitter.com/ncg1fWBKN0
Musk also announced the development of a second supercomputing cluster, which will be launched soon. The new facility will initially operate with a batch of 550,000 Nvidia GB200 and GB300 chips, marking a significant leap in AI infrastructure.
The $13 billion question: What xAI will spend in 2025
Estimates in the trade press suggest xAI may spend around $13 billion in 2025, indicating the financial scale required to stay competitive in the AI race. The company's efforts to scale AI hardware and computing capabilities come in direct competition with firms like OpenAI, Alphabet, and China's DeepSeek.
The @xAI goal is 50 million in units of H100 equivalent-AI compute (but much better power-efficiency) online within 5 years
In July, The Financial Times had reported that xAI was seeking new funding, with a valuation between $170 billion and $200 billion. Musk denied this at the time, stating, 'We have more than enough capital.'
Musk's AI hardware vision anchored on Nvidia chips
Musk has also stated that xAI's long-term vision involves deploying the equivalent of 50 million Nvidia H100 GPUs over the next five years. The H100 is currently an industry standard for AI training and inference. Musk's plan also includes achieving better power efficiency than Nvidia's flagship chips, suggesting an effort to both scale and innovate.
While Nvidia remains central to AI compute globally, Musk's entry and investment reinforce just how high the stakes are in the AI infrastructure race.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
The Print
an hour ago
- The Print
Elon Musk's Starlink network suffers rare global outage
Starlink, which has more than 6 million users across roughly 140 countries and territories, later acknowledged the outage on its X account and said 'we are actively implementing a solution.' Users in the U.S. and Europe began experiencing the outage at around 3 p.m. EDT (1900 GMT), according to Downdetector, a crowdsourced outage tracker that said as many as 61,000 user reports to the site were made. (Reuters) -SpaceX's Starlink suffered one of its biggest international outages on Thursday when an internal software failure knocked tens of thousands of users offline, a rare disruption for Elon Musk's powerful satellite internet system. Starlink service mostly resumed after 2.5 hours, Michael Nicolls, Starlink vice president of Starlink Engineering, wrote on X. 'The outage was due to failure of key internal software services that operate the core network,' Nicolls said, apologizing for the disruption and vowing to find its root cause. Musk had also apologized: 'Sorry for the outage. SpaceX will remedy root cause to ensure it doesn't happen again,' the SpaceX CEO wrote on X. The outage was a rare hiccup for SpaceX's most commercially sensitive business that had experts speculating whether the service, known for its resilience and rapid growth, was beset by a glitch, a botched software update or even a cyberattack. Doug Madory, an expert at the internet analysis firm Kentik, said the outage was global and that such a sweeping interruption was unusual. 'This is likely the longest outage ever for Starlink, at least while it became a major service provider,' Madory said. As Starlink gained more users, SpaceX has focused heavily in recent months on updating its network to accommodate demands for higher speed and bandwidth. The company in a partnership with T-Mobile is also expanding the constellation with larger, more powerful satellites to offer direct-to-cell text messaging services, a line of business in which mobile phone users can send emergency text messages through the network in rural areas. SpaceX has launched more than 8,000 Starlink satellites since 2020, building a uniquely distributed network in low-Earth orbit that has attracted intense demand from militaries, transportation industries and consumers in rural areas with poor access to traditional, fiber-based internet. 'I'd speculate this is a bad software update, not entirely dissimilar to the CrowdStrike mess with Windows last year, or a cyberattack,' said Gregory Falco, director of a space and cybersecurity laboratory at Cornell University. An update to CrowdStrike's widely used cybersecurity software led to worldwide flight cancellations and impacted industries around the globe in July last year. The outage disrupted internet services, affecting 8.5 million Microsoft Windows devices. It was unclear whether Thursday's outage affected SpaceX's other satellite-based services that rely on the Starlink network. Starshield, the company's military satellite business unit, has billions of dollars' worth of contracts with the Pentagon and U.S. intelligence agencies. (Reporting by Juby Babu in Mexico City, and Raphael Satter and Joey Roulette in Washington; Editing by Anil D'Silva and Matthew Lewis) Disclaimer: This report is auto generated from the Reuters news service. ThePrint holds no responsibility for its content.
Time of India
an hour ago
- Time of India
Nvidia places huge order of its most powerful AI chip in China but why it may not meet demands
Nvidia has reportedly placed a significant order for H20 chipsets with contract manufacturer Taiwan Semiconductor Manufacturing Company (TSMC) last week. News agency Reuters report that this move signals a change in strategy, citing sources as saying that strong Chinese demand has prompted Nvidia to increase production rather than solely rely on its existing stockpile. The report said that Nvidia has ordered TSMC to manufacture 300,000 H20 chips that will add to the US chipmaker's current inventory of 600,000 to 700,000 H20 chips. For context, US research firm SemiAnalysis estimates Nvidia sold approximately 1 million H20 chips in 2024. Nvidia has also reportedly requested that Chinese companies interested in purchasing the H20 chips submit new documentation, including client order volume forecasts. Why Nvidia's latest H20 AI chip order for China is important Firstly, this order comes after the Trump administration reversed an effective April ban earlier this month, allowing Nvidia to resume sales of its H20 graphics processing units (GPUs) to China. The initial ban was implemented to restrict advanced AI chip access to Chinese entities, citing national security concerns. by Taboola by Taboola Sponsored Links Sponsored Links Promoted Links Promoted Links You May Like Pierce Brosnan's Wife Lost 120 Pounds - This Is Her Now Undo Crucially, Nvidia requires export licenses from the US government to ship these H20 chips. While the company stated in mid-July it had received assurances from authorities regarding expedited approvals, the US Commerce Department has yet to officially approve these licenses, according to multiple sources cited by Reuters. Nvidia specifically developed the H20 for the Chinese market following broader US export restrictions imposed on its more powerful AI chipsets like the H100 and the new Blackwell series, which are sold in other global markets. Nvidia CEO Jensen Huang had previously stated this month during a visit to Beijing that the decision to restart H20 production would hinge on order volumes, noting that reactivating the supply chain would take about nine months. Prior to the April ban, major Chinese tech firms including Tencent, ByteDance, and Alibaba significantly increased their H20 orders, deploying them for cost-effective AI models like DeepSeek's, as well as their own. AI Masterclass for Students. Upskill Young Ones Today!– Join Now
Time of India
2 hours ago
- Time of India
Will AI make learning more personalised or passive for students? Jensen Huang and Ramine Tinati offer opposing views
As artificial intelligence continues to redefine how work and knowledge are distributed, education stands at a pivotal crossroad. Will AI help students learn in a more customised, efficient manner, or will it encourage disengagement by doing too much of the intellectual heavy lifting? The answer, according to two of today's tech voices, depends on how we design learning systems and how we rethink the purpose of education itself. During the Fortune Brainstorm AI conference in Singapore, where Ramine Tinati, lead at Accenture's APAC Center for Advanced AI, expressed caution about AI's impact on productivity and learning. Earlier this year, Nvidia CEO Jensen Huang offered a contrasting take on the relationship between AI and cognitive growth during his appearance on CNN's Fareed Zakaria GPS . Together, their perspectives highlight a growing divide in how experts view AI's role in student learning: as either an enabler of deeper thinking or a tool that risks fostering passivity. Personalised tools or automated shortcuts? Speaking in Singapore, Tinati questioned the assumption that faster task completion equals improved productivity or learning. 'If you give employees a tool to do things faster, they do it faster. But are they more productive? Probably not, because they do it faster and then go for coffee breaks,' he said, referring to the corporate world. The parallel in education is clear: AI tools may help students complete assignments quickly, but does that equate to deeper understanding? Tinati warned that meaningful progress requires more than bolting AI onto existing structures. by Taboola by Taboola Sponsored Links Sponsored Links Promoted Links Promoted Links You May Like Sintering Furnace Across International Get Quote Undo For schools and universities, this raises critical questions about whether classroom workflows, assessment methods, and pedagogy are being redesigned to take full advantage of AI or if technology is merely being layered on top of outdated systems. His comments suggest that without a holistic reinvention of how education is delivered, students may become increasingly passive learners, reliant on AI-generated summaries, solutions, and answers instead of developing analytical and creative skills. Intelligence amplifier or crutch? Jensen Huang, on the other hand, sees AI not as a shortcut but as an amplifier. Rejecting concerns raised by a Massachusetts Institute of Technology (MIT) study that suggested frequent AI use may impair cognitive function, the Nvidia CEO offered a personal counterpoint. Huang said he uses AI 'literally every single day,' on Fareed Zakaria GPS in July 2025 and says his 'cognitive skills are actually advancing.' His stance reflects the idea that AI, when used intentionally, can deepen intellectual engagement by freeing up mental bandwidth. Instead of spending time on rote memorisation or repetitive tasks, students could spend more time asking deeper questions, synthesising ideas, or applying knowledge in creative contexts. A shift in educational philosophy As schools and universities integrate generative AI tools into classrooms, the tension between automation and agency becomes more urgent. Adaptive learning platforms, AI tutoring bots, and automated grading systems offer the promise of scale and personalisation, but they also risk removing the friction that is often essential to deep learning. Educators remain divided. Some see AI as a valuable co-pilot, helping differentiate instruction and offer real-time feedback. Others worry that over-reliance on AI will erode students' critical thinking and reduce opportunities for productive struggle, an essential part of learning. What both camps agree on, however, is that AI's impact on education will be shaped less by the technology itself and more by the values, structures, and pedagogy surrounding its use. Tinati's call to 'reinvent the work' resonates with education leaders who argue that merely digitising textbooks or automating homework checks is not enough. True transformation requires rethinking how knowledge is acquired, applied, and assessed. For some, this means designing learning environments where AI acts as a guide, not a substitute. For others, it means slowing down the rush to automate in favour of deeper dialogue around learning outcomes. What's ahead As AI becomes more embedded in classrooms and curricula, its effects will likely remain contested. The divide between those who view it as a tool for liberation and those who see it as a threat to student autonomy reflects a broader uncertainty about the future of learning. For now, the debate between voices like Jensen Huang and Ramine Tinati underscores a vital question for educators and policymakers alike: Are we shaping AI to serve the goals of education or allowing it to reshape education without reflection? TOI Education is on WhatsApp now. Follow us here . Ready to navigate global policies? Secure your overseas future. Get expert guidance now!
