Connect with us

Published

on

Meta has built custom computer chips to help with its artificial intelligence and video-processing tasks, and is talking about them in public for the first time.

The social networking giant disclosed its internal silicon chip projects for the first time to reporters earlier this week, ahead of a Thursday virtual event discussing its AI technical infrastructure investments.

related investing news

Nvidia's stock could rise fivefold in 10 years on A.I. trend, says fund manager

CNBC Pro

Investors have been closely watching Meta’s investments into AI and related data center hardware as the company embarks on a “year of efficiency” that includes at least 21,000 layoffs and major cost cutting.

Although it’s expensive for a company to design and build its own computer chips, vice president of infrastructure Alexis Bjorlin told CNBC that Meta believes that the improved performance will justify the investment. The company has also been overhauling its data center designs to focus more on energy-efficient techniques, like liquid cooling, to reduce excess heat.

One of the new computer chips, the Meta Scalable Video Processor (MSVP), is used to process and transmit video to users while cutting down on energy requirements. Bjorlin said “there was nothing commercially available” that could handle the task of processing and delivering 4 billion videos a day as efficiently as Meta wanted.

The other processor is the first in the company’s Meta Training and Inference Accelerator (MTIA) family of chips intended to help with various AI-specific tasks. The new MTIA chip specifically handles “inference,” which is when an already-trained AI model makes a prediction or takes an action.

Bjorlin said that the new AI inference chip helps power some of Meta’s recommendation algorithms used to show content and ads in people’s news feeds. She declined to answer who is manufacturing the chip, but a blog post said that the processor is “fabricated in TSMC 7nm process,” indicating that chip-giant Taiwan Semiconductor Manufacturing is producing the technology.

She said that Meta has a “multi-generational roadmap” for its family of AI chips that include processors used for the task of training AI models, but declined to offer details beyond the new inference chip. Reuters previously reported that Meta cancelled one AI inference chip project and started another that was supposed to roll out around 2025, but Bjorlin declined to comment on that report.

Because Meta isn’t in the business of selling cloud computing services like companies including Google-parent Alphabet or Microsoft, the company didn’t feel compelled to publicly talk about its internal data center chip projects, she said.

“If you look at we’re sharing—our first two chips that we developed—it’s definitely giving a little bit of a view into what are we doing internally,” Bjorlin said. “We haven’t had to advertise this, and we don’t need to advertise this, but you know, the world is interested.”

Meta vice president of engineering Aparna Ramani said the company’s new hardware was developed to work effectively with its home-grown PyTorch software, which has become one of the most popular tools used by third-party developers to create AI apps.

The new hardware will eventually be used to power tasks related to the metaverse, such as virtual reality and augmented reality, as well as the burgeoning field of generative AI, which generally refers to AI software that can create, compelling text, images, and videos.

Ramani also said that Meta has developed a generative AI-powered coding assistant for the company’s developers to help them more easily create and operate software. The new assistant is similar to Microsoft’s GitHub Copilot tool that it released in 2021 with help from the AI startup OpenAI.

In addition, Meta said it completed the second-phase buildout, or the final buildout, of its supercomputer dubbed Research SuperCluster (RSC), which the company detailed last year. Meta used the supercomputer, which contains 16,000 Nvidia A100 GPUs, to train the company’s LLaMA language model, among other uses.

Ramani said that Meta continues to act on its belief that it should contribute to open-source technologies and AI research in order to push the field of technology. The company has disclosed that its biggest LLaMA language model, LLaMA 65B, contains 65 billion parameters and was trained on 1.4 trillion tokens, which refers to the data used for AI training.

Companies like OpenAI and Google have not publicly disclosed similar metrics for their competing large language models, although CNBC reported this week that Google’s PaLM 2 model was trained on 3.6 trillion tokens and contains 340 billion parameters.

Unlike other tech companies, Meta released its LLaMA language model to researchers so they can learn from the technology. However, the LlaMA language model was then leaked to the wider public, leading to many developers building apps incorporating the technology.

Ramani said that Meta is “still thinking through all of our open source collaborations, and certainly, I want to reiterate that our philosophy is still open science and cross collaboration.”

Watch: A.I. is a big driver of sentiment for big tech

Continue Reading

Technology

SpaceX’s Starship explodes during routine test in Texas

Published

on

By

SpaceX's Starship explodes during routine test in Texas

A SpaceX Starship is seen in Boca Chica, Texas in 2023.

Patrick T. Fallon | Afp | Getty Images

A SpaceX Starship rocket on Wednesday exploded at the Starbase facility in Texas during routine testing in preparation for a launch flight, according to local authorities and live stream footage.

The rocket “experienced a major anomaly while on a test stand at Starbase” at 11 p.m. local time, SpaceX said on social media, noting “a safety clear area around the site was maintained throughout the operation and all personnel are safe and accounted for.”

Local authorities said that Starship “suffered a catastrophic failure and exploded,” with no injuries reported at the time of writing and an investigation is now underway. Live stream footage of Starbase showed the rocket burst into flame, shooting a large fireball into the sky.

Another Starship launch was expected to take place by the end of this month.

It’s been a tempestuous ride for Elon Musk’s mammoth Starship, after three flight launch attempts devolved in fiery glory and air-traffic stopping debris this year to date. Notably, the rocket model has taken off successfully in previous instances, but its vast scale — standing 120 meters (394 feet) tall when factoring in the Super Heavy booster — has raised concerns over its overall reliability and requirements for orbital refueling once in flight.

Yet Musk has clinched his hopes on Starship as the key vehicle for both NASA’s third and fourth Artemis missions — part of a broader plan to return humans to the Moon — due to take place over 2027-2028. The rocket is also set to play a role in launching the Starlab private space station in the transition to commercial space orbiting labs once the International Space Station retires after 2030.

Critically, Starship is also central to Musk’s — and former ally U.S. President Donald Trump’s — broader ambitions to colonize Mars. The rocket is set to ferry Optimus robots to the red planet by the end of 2026, with Musk in March saying, “If those landings go well, then human landings may start as soon as 2029, although 2031 is more likely.”

Subscribe to CNBC’s Investing in Space weekly newsletter

Continue Reading

Technology

Apple looking to make ‘premium’ priced folding iPhones starting next year, analyst says

Published

on

By

Apple looking to make 'premium' priced folding iPhones starting next year, analyst says

People look at iPhones at the Apple Fifth Avenue store in New York City on May 23, 2025.

Adam Gray | Reuters

Apple has plans to make a folding iPhone starting next year, reliable analyst Ming-Chi Kuo said on Wednesday.

Kuo said Apple’s folding phone could have a display made by Samsung Display, which is planning to produce as many as eight million foldable panels for the device next year. However, other components haven’t been finalized, including the device’s hinge, Kuo wrote. He expects it to have “premium pricing.”

Kuo is an analyst for TF International Securities, and focuses on the Asian electronics supply chain and often discusses Apple products before they’re launched.

He wrote in a post on social media site X that Apple’s plans for the foldable iPhone aren’t locked in yet and are subject to change. Apple did not respond to CNBC’s request for comment.

Apple’s iPhone makes up over half of Apple’s business and remains an incredibly profitable product, accounting for $201 billion in sales in the company’s fiscal 2024. But iPhone revenue peaked in 2022, and Apple is constantly looking for ways to attract new customers and convince its current customers to upgrade to more expensive devices.

The Flex S is another concept device Samsung showed off at MWC. It folds in a more zigzag-like way to make an “S” shape.

Ryan Browne | CNBC

Several of Apple’s rivals, including Huawei and Samsung, have been releasing folding smartphones since 2019.

The devices promise the screen size of a tablet in a format that can be stored in pants pockets. But folding phones still have hardware issues, including creases in the display where it is folded.

Folding phones also have yet to prove they drive significant demand after the novelty wears off.

Research firm TrendForce said last year that only 1.5% of all smartphones sold can fold. Counterpoint, another research firm tracking smartphone sales, said earlier this year that the folding market only grew about 3% in 2024 and is expected to shrink in 2025.

Read more CNBC tech news

Continue Reading

Technology

Scale AI not ‘winding down’ following Meta deal, interim CEO tells employees and customers

Published

on

By

Scale AI not 'winding down' following Meta deal, interim CEO tells employees and customers

FILE PHOTO: Jason Droege speaks at the WSJTECH live conference in Laguna Beach, California, U.S. October 22, 2019.

Mike Blake | Reuters

Scale AI’s Interim CEO Jason Droege said in a memo on Wednesday that the artificial intelligence startup is not changing course following Meta’s multibillion-dollar investment in the company last week.

“Unlike some other recent tech deals you might have heard about in the AI space, this is not a pivot or a winding down,” Droege wrote in a post directed at customers, employees and investors.

Meta has a 49% stake in Scale after its $14.3 billion investment, though the social media company will not have any voting power. Scale AI’s founder Alexandr Wang, along with a small number of other Scale employees, will join Meta as part of the agreement.

“Scale remains, unequivocally, an independent company,” Droege wrote. “This deal rewards many of the people who helped build Scale into what it is today, but more importantly to me, it’s also a validation of the course we’re on.”

Scale AI appointed Droege, the company’s chief strategy officer, to serve as its interim chief executive following the deal. Droege wrote that Scale AI is still “a well-resourced company” that has “multiple promising lines of business.”

Founded in 2016, Scale AI rose to prominence by helping major tech companies like OpenAI, Google and Microsoft prepare data they use to train cutting-edge AI models. Meta has been one of Scale AI’s biggest customers.

Droege said the company is “not slowing down” and remains committed to its data and application business units. Scale will also continue to be model agnostic, he added.

“The need for high-quality data for AI models remains significant, and with the largest network of experts training AI, we are set up well to help model builders keep pushing the frontier of what’s possible,” Droege wrote.

But some of Scale AI’s tech customers may be having doubts.

OpenAI confirmed to CNBC on Wednesday that it has been wrapping up its work with Scale AI over the past six to 12 months. The company said it’s looking to work with other data providers that have kept pace with innovation, and that its decision to wind down its work with Scale wasn’t influenced by the Meta partnership.

Google is also reportedly cutting ties with Scale following the company’s deal with Meta, according to a report from Reuters. Google declined to comment.

WATCH: Scale AI CEO departs for Meta in Zuckerberg’s latest AI gambit

Scale AI CEO departs for Meta in Zuckerberg’s latest AI gambit

Continue Reading

Trending