| |  | Feb 26, 2026 | | | | | Supported by |  | | | | | | Happy Thursday! Amazon's investment in OpenAI could hinge on IPO or AGI. Nvidia's revenue grew 73% to $68.1 billion in the three months ended in January. Bill Gates apologiezes for his relationship with Epstein to his foundation staff.
| | | | As Amazon weighs a major equity investment in OpenAI, the ultimate size of its commitment could hinge on whether OpenAI goes public or achieves a loosely defined milestone known as artificial general intelligence, The Information reported Wednesday. The proposed Amazon investment is part of OpenAI's current funding round, which could top $100 billion at a valuation of $730 billion before the financing. Under terms still being negotiated, Amazon would invest $15 billion upfront, with an additional $35 billion tied to the achievement of the specific milestones. OpenAI's existing partnership with Microsoft is influencing the structure of the pending Amazon deal, The Information reported. Microsoft currently has the exclusive right to run OpenAI models on its cloud business Azure until OpenAI reaches AGI. | | | | Nvidia's revenue rose 73% to $68.1 billion in the three months ended in January, 11 percentage points higher than the remarkable growth it reported in the November quarter. It also projected even faster revenue growth—nearly 77%—in the current fiscal quarter. The results shows Nvidia, the world's most valuable company, is still in a league of its own in selling the primary hardware that powers AI and its performance is improving even as its stock gets cheaper on a multiple of its expected earnings. The stock is trading at 27 times forward earnings, according to S&P, down from 32 times forward earnings as of three months ago. As always, one of the main long-term concerns about the stock is that a handful of customers make up a majority of Nvidia's revenue, and their increased spending on Nvidia chips is eating into their cash balances. Nvidia CEO Jensen Huang responded to that concern by saying he believed the spending on AI chips was producing profitable revenue among Nvidia customers. Nvidia shares rose 3% in after-hours trading and was up around 4% so far this year before the close of normal trading Wednesday. Advanced Micro Devices, whose revenues are more than six times smaller than Nvidia's reported revenue growth of 34% in the December quarter. Nvidia generated $216 billion in the fiscal year that ended in January, with $194 billion of that total coming from its data center business, which includes its AI chip and networking sales. In the January quarter, it generated $11 billion in networking revenue, up 263% from a year earlier, as customers increasingly used its NVLink product to connect their AI server clusters. Nvidia's gross margin improved 2 percentage points in the quarter, compared to the same period a year earlier, and free cash flow more than doubled to nearly $35 billion. In addition to dividiends and share buybacks, Nvidia has been making large investments in some of its biggest customers, and recently made its largest acquisition-like deal when it licensed technology from AI chip designer Groq. Nvidia said Wednesday it paid $13 billion so far, with $4 billion in additional costs related to the deal to be paid within a year. The company also has been considering making a $30 billion equity investment in OpenAI, The Information has reported. "We are finalizing an investment and partnership agreement with OpenAI. There is no assurance that we will enter into an investment and partnership agreement with OpenAI or that a transaction will be completed," Nvidia said in its annual filing. | | | | Bill Gates told staff of his charitable foundation that he's sorry for maintaining a relationship with Jeffrey Epstein, but said he didn't participate in any of Epstein's crimes, the Wall Street Journal reported Wednesday. Files recently released by the Justice Department indicated that Gates met several times with Epstein beginning in 2011, years after the disgraced financier was convicted of soliciting a minor for prosecution, and photos released as part of the files show Gates alongside women whose faces are redacted. Gates told staff that he failed to adequately check Epstein's background and that the women he was pictured with were Epstein's assistants, not his victims, according to the report. The released files also show Epstein sending emails that allude to Gates' extramarital affairs. Gates reportedly told staff this week that he did have affairs with two Russian women while married to his then-wife Melinda, but that he didn't meet either of the women through Epstein. | | | | Salesforce shares fell more than 5% in after-hours trading as the enterprise software firm said Wednesday that revenue in the January fiscal quarter rose 12% to $11.2 billion, including sales from Informatica, a data management firm it acquired late last year. Excluding revenue from Informatica, Salesforce revenue climbed 8% in the period, 1 percentage point lower than its growth in the prior quarter. Salesforce projected revenue would climb 8% to 9% in the April quarter, excluding Informatica sales. Investors have been waiting for Salesforce to show that its internal AI products, namely Agentforce, will boost overall revenue. On the bright side, Agentforce annual recurring revenue—typically defined as the current monthly subscription multiplied by 12—rose nearly three times to $800 million. "Agentic AI is a tailwind for our business," CEO Marc Benioff said in a statement. Salesforce stock had fallen nearly 28% so far this year before the market closed Wednesday, as software equities have been hammered by investor fears about the impact of AI on sales growth in the sector. | | | | OpenAI has hired a high-profile AI researcher at Meta Platforms who had joined the company from Apple about seven months ago, a spokesperson from OpenAI confirmed. Ruoming Pang, who led AI infrastructure for Meta Superintelligence Labs, which is developing new AI models, left the company last week and has joined OpenAI. Pang, who was aggressively recruited by OpenAI over several months, told colleagues he was happy at Meta and that its infrastructure was in good shape, according to a person with direct knowledge of the matter. A spokesperson for Meta didn't have a comment. Pang previously led model development at Apple and joined Meta last year for a pay package reportedly worth more than $200 million over several years and contingent on meeting certain milestones. The move was a blow to Apple, which has been struggling to develop AI models and features for its devices. Pang's move to OpenAI follows a wave of departures from Meta's AI teams over the past year, though Meta has also hired numerous people from OpenAI and other AI competitors. Mat Velloso, who led product for developer platforms at Meta Superintelligence Labs, also departed after a short stint. He joined Meta's AI organization from Google DeepMind in July and announced his departure on LinkedIn last week. And Yann LeCun, a longtime chief AI scientist at Meta, left last year. | | | | Bob McGrew, the former chief research officer at OpenAI and an adviser to Thinking Machines Lab, is starting a new company that will build AI-powered software for manufacturing, according to a person with knowledge of the company. McGrew is cofounding the startup with Augustus Odena, a well known AI researcher whose personal website says he invented the technique underlying reasoning models and who previously co-founded AI agent startup Adept, along with former Palantir engineers Jakob Frick and Alex Mark, the person said. The startup is called Arda, a reference to the setting of Lord of the Rings. The startup has begun discussions with investors about raising capital, the person said, though the amount of funding and valuation couldn't be learned. Arda is yet another example of an AI startup founded by former high-profile researchers and executives from the major AI labs. Sometimes called neolabs, these startups aim to leverage new approaches to developing AI models and sometimes focus on industries like manufacturing they say major developers like OpenAI and Anthropic may have overlooked. Jerry Tworek, a senior researcher at OpenAI, left the company earlier this year to start his own neolab, which has been looking to raise up to $1 billion in funding. | | | | Nvidia said it has provided $3.5 billion in guarantees to companies leasing land, power and data center facilities, four times the amount it disclosed the previous quarter. The guarantees could help companies without credit ratings, such as AI cloud startups, secure coveted power resources and debt financing to build data centers. Nvidia said it provided the guarantees to "early-stage companies," without disclosing their identities. The chip company said it could take over the underlying leases if the companies defaulted and find another tenant or repurpose the facilities for its own use. Nvidia also said Wednesday it had been asked to "offer financing arrangements to support our customers' and partners' buildout of data center infrastructure" but that it hadn't done any of those types of deals. | | | | DeepSeek didn't give Nvidia or Advanced Micro Devices early access to its upcoming flagship AI model for performance optimization, Reuters reported, citing two people familiar with the matter. DeepSeek, meanwhile, granted domestic chip companies, including Huawei Technologies, a head start of several weeks so they can tune the software for their processors, according to Reuters. AI developers usually share pre-release versions of new models with chip companies to ensure the software runs efficiently on different processors. It is unclear why DeepSeek chose to exclude the U.S. firms this time. In December, The Information reported that DeepSeek's new model is trained on Nvidia's most advanced Blackwell chips that were smuggled into China, even though the U.S. has banned the export of such chips to China. A senior Trump administration official told Reuters earlier this week that DeepSeek trained its latest model on Blackwell chips using a cluster in mainland China. | | | | General Atlantic is selling part of its stake in TikTok-parent ByteDance at a $550 billion valuation, according to a report from Reuters. A sale at this level would be a marked jump from the $330 billion the company was valued at in a share buyback last year, the report says. In January, the U.S. government approved new ownership of a joint venture called TikTok U.S. in order to comply with a U.S. law requiring TikTok's U.S. operations separate from China's ByteDance, General Atlantic declined to comment on the report. General Atlantic first invested in Bytedance in 2017, when it was valued about $20 billion, and has a board seat. As of early 2024, when ByteDance was valued at about $200 billion, the investment firm hadn't sold any of its stake, a senior executive previously said. The report also says that Bytedance hit $48 billion in revenue in the second quarter of last year. | | | | Shares of Snowflake dropped more than 2% after its fourth-quarter earnings report after the cloud database provider forecast slower revenue growth for its current quarter and fiscal year compared to last year. Still, investors seem to be less concerned about the potential threat AI poses to Snowflake's business than they are about its impact on application providers like Salesforce, ServiceNow and Workday, whose shares are all down significantly since the start of the year. The big question is how much AI products—which include the data analytics tool Snowflake Intelligence and AI coding agent Cortex Code—will contribute to Snowflake's future growth. Snowflake revealed in December that its AI products hit $100 million in annual recurring revenue, but executives didn't provide an updated figure on the earnings call. (The Information was first to report on the sales target last April.) On an earnings call, Snowflake CFO Brian Robins acknowledged that the profit margins for AI products aren't as high as they are for the company's older products. However, Snowflake has responded by using AI to run its business more efficiently, and as a result, laid off around 200 employees in the fourth quarter while only hiring 37 new employees, Robins said on the call. For its January quarter, Snowflake's product revenue (the vast majority of its overall revenue) was $1.23 billion, up 30% from last year's quarter. For its current quarter, Snowflake is expecting product revenue of $1.262 billion and $1.267 billion, up 27% from last year. Snowflake is expecting revenue of $5.66 billion for its fiscal year ending in January, up 27% from last year, compared to $4.5 billion and 29% year-over-year growth last year. | | | | | Popular articles
By Sri Muppidi and Stephanie Palazzolo
| | | | | Opportunities Empower your teams to stay ahead of market trends with the most trusted tech journalism.
Learn more
Reach The Information's influential audience with your message. Connect with our team | | | | | |
0 comentários:
Postar um comentário