| |  | Mar 27, 2026 | | | | | Supported by |  | | | | | | | | | Anthropic is discussing an initial public offering that could happen as soon as the fourth quarter of this year, The Information reported Thursday. Bankers vying to take the company public have said it could raise as much as $60 billion, the report said. Meanwhile, Anthropic on Thursday said it was preparing to release Claude Mythos, its next flagship AI model. An Anthropic spokesperson told Fortune that Mythos represented "a step change" in AI performance and was "the most capable we've built to date." The statement came after Fortune reported on an unreleased blog post from Anthropic, which referenced Mythos and said the model carried "unprecedented cybersecurity risks." (Companies usually evaluate prereleased models for such risks and can tweak them to lower the risk.) The unpublished post said select customers are currently testing Mythos, but it isn't clear if that has happened yet. The post also said Mythos was expensive to run, the report said. Archival OpenAI is trying to release its own new flagship model, codenamed Spud, after completing so-called pretraining work on it, The Information reported earlier this week. The unreleased Anthropic blog post also described a new model "tier" called Capybara, which Anthropic says is "larger and more intelligent than our [flagship Claude] Opus models—which were, until now, our most powerful," according to Fortune. It isn't clear if there is a difference between Mythos and Capybara or if the names are related to the same underlying model. "Compared to our previous best model, Claude Opus 4.6, Capybara gets dramatically higher scores on tests of software coding, academic reasoning, and cybersecurity, among others," the company said in the blog, the Fortune report said. | | | | A federal judge in California granted Anthropic a preliminary injunction halting the Pentagon's designation of the AI startup as a supply chain risk while Anthropic's lawsuit against the Department of Defense proceeds. In her order, Judge Rita Lin of the Northern District of California wrote that the supply chain risk designation was "likely both contrary to law and arbitrary and capricious." While the injunction itself won't necessarily result in federal agencies resuming work with Anthropic, it could reassure other current or potential Anthropic customers. Lin wrote that the order only restored the status quo prior to Feb. 27, when federal agencies were still free to choose whether or not to use Anthropic's AI. The injunction will go into effect after a seven-day administrative stay giving the Pentagon a chance to appeal, a common step in major decisions. Once in effect, the injunction will require agencies to cease implementation of President Donald Trump's Feb. 27 post instructing federal agencies to stop using Anthropic and the March 3 letter to Anthropic officially designating it a supply chain risk. The Pentagon additionally designated Anthropic a supply chain risk under a separate statute, which Anthropic is also challenging in court in Washington, D.C. Anthropic is awaiting a decision in that case. | | | | OpenAI has surpassed $100 million in annualized revenue from its ChatGPT ads business, six weeks after the pilot was announced, according to a spokesperson. That revenue has been generated from the less than 20% of U.S.-based ChatGPT Free and Go users who are shown ads on a daily basis today, though around 85% of Free and Go users are eligible to see ads, the spokesperson said. OpenAI has expanded to more than 600 advertisers, and is on track to launch self-serve advertiser access in April, they said. The company is currently focused on improving ad relevance, with less than 7% of ads rated by users as "low relevance," as well as ensuring that ads don't negatively impact user trust, the spokesperson said. The company announced earlier this week that it had brought on former Meta ad executive Dave Dugan to lead ads sales. It's now exploring expanding ads into other geographic regions including Canada, Australia and New Zealand, the spokesperson said. Ads represent an important source of revenue as the company gears up to go public. Executives have told investors that the company expects to generate more than $17 billion from consumers using ChatGPT in 2026, including making money from free users through advertisements. | | | | Microsoft recently suspended new hiring across major divisions, including its Azure cloud and North American sales groups, as the company seeks to offset its investments in artificial intelligence infrastructure, The Information reported. Executives have instructed managers in these sprawling divisions to freeze all potential hires who do not already have job offers, prioritizing cost restraint and gross margin improvements before the fiscal year ends in June. The targeted hiring freeze highlights growing investor pressure on Microsoft and other cloud and software firms, which are also laying off staff or slowing their hiring. Some senior Microsoft executives believe the company's headcount won't grow in the coming years, the report said. | | | | Elon Musk's X laid off its chief marketing officer and more than 20 other staff in non-engineering roles in recent weeks, the Wall Street Journal reported. The company has also brought in Jon Shulkin, a longtime Musk investor and ally, to work as chief revenue officer as the site tries to cut costs and boost revenue, the Journal reported. X was acquired by Musk's AI firm, xAI, last year. And in turn, xAI was acquired by SpaceX in February. Now, SpaceX is planning to go public later this year, but X, the company formerly known as Twitter, is unlikely to play a major role in its pitch to IPO investors, The Information has reported. In one effort to grow its business, X plans to launch an initial version of its payments service, X Money, in the coming weeks, Musk has said. But the service has been delayed for years, partly from problems securing money transmitter licenses in several U.S. states, including New York, The Information has reported. | | | | David Sacks said he has stepped down as the Trump administration's AI and crypto czar, confirming Thursday that he exhausted his 130-day limit as a special government employee. In an interview with Bloomberg Television, the venture capitalist, who took the role at the start of Trump's second term, said he will continue to advise Trump as a co-chair of the President's Council of Advisors on Science and Technology. Earlier this week, Trump named Sacks to the panel, which also includes Nvidia's Jensen Huang, Meta Platforms' Mark Zuckerberg, and Oracle's Larry Ellison. "As co-chair of PCAST, I can now make recommendations on not just AI but an expanded range of technology topics," Sacks said in the interview. Sacks is a partner at Craft Ventures and was a prominent donor to Trump's presidential campaign in 2024. | | | | OpenAI has put plans to allow its ChatGPT chatbot to generate erotica on hold indefinitely, according to a report from the Financial Times. OpenAI CEO Sam Altman first teased "adult mode" in ChatGPT last October, saying in a post on X that the release would be part of OpenAI's "treat adult users like adults" principle. In more recent months, however, the company has cut back on side projects like its video generation product Sora and instant checkout initiative, as it looks to focus more on its core products for consumers and businesses. OpenAI declined to comment on its plans for ChatGPT "adult mode." | | | | Aligned Data Centers, the developer being purchased by a consortium including BlackRock and Abu Dhabi's MGX, raised $2.6 billion in debt to help it build more AI computing facilities in the U.S. Aligned said the lenders included "sophisticated insurance, pension fund and other institutional capital," without providing the names. The company said it structured the debt so that it would count as investment-grade capital for insurance funds. The consortium buying Aligned paid about $40 billion for the developer, including debt, in the largest data center sale to date. Aligned said the transaction is scheduled to close in the first half of this year. Aligned also last year expanded a credit facility with Blackstone's credit and insurance arm to more than $1 billion in potential borrowing. | | | | Gadi Hutt, a director of product and customer engineering in Amazon's server-chip design unit, Annapurna Labs, has left the company, according to two people with direct knowledge of the move. Hutt was an early employee at Annapurna Labs, the Israeli chip startup Amazon acquired in 2015 for $350 million. He has been one of the unit's most visible spokespeople for Trainium, Amazon's chips for training and running AI models, which aim to be more cost-effective than graphic processing units from Nvidia. On an earnings call in November, Amazon CEO Andy Jassy said Trainium is a multibillion-dollar business that grew 150% in the third quarter last year. That's largely due to Anthropic's rentals of the chips, The Information reported. More recently, Amazon struck a deal with Cerebras to incorporate that company's AI chip into Trainium server systems. The collaboration aims to speed up the performance of running AI models, otherwise known as inference. As of late 2024, Hutt said the company was primarily focused on getting large cloud customers to use Trainium chips and expected them to be broadly adopted. "This is a marathon, not a sprint," he said. Hutt is the second Annapurna leader to depart in the past seven months. Last August, Rami Sinno, director of engineering at Annapurna Labs, left to join chip designer Arm Holdings to help lead that company's efforts to develop its own chips. An Amazon spokesperson didn't have a comment on the record. Hutt didn't respond to requests for comment. | | | | China's largest chipmaker, Semiconductor Manufacturing International Corp., has been supplying chipmaking tools to Iran's military for roughly a year, Reuters reported, citing two U.S. officials. The tools were sent to Iran's military-industrial complex and could be used to produce chips for a broad range of electronics, according to Reuters. It remains unclear whether any of the equipment is of U.S. origin, a detail that would constitute a direct violation of American sanctions. SMIC has been on a U.S. trade blacklist since 2020 over alleged ties to China's military, which both SMIC and the Chinese government have denied. The allegations add fresh friction to U.S.-China relations as Washington wages war against Tehran alongside Israel. China has not publicly taken a stance in the U.S.-Iran conflict. | | | | | Popular articles By Stephanie Palazzolo and Amir Efrati
By Katie Roof and Valida Pau
| | | | | 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