Cars with no steering wheels… great.
THE DOWNLOAD
Meta’s chip shot
Meta
TL;DR: Meta is sprinting into the custom chip race, announcing plans to roll out four new in-house AI processors by 2027. The move mirrors a broader push across Big Tech to control more of the AI hardware stack as computing costs surge. But designing chips isn’t easy, and even if Meta pulls it off, powering AI features for its 3.58 billion daily users will still require a lot of third-party GPUs. Safe to say, this move doesn’t put Jensen Huang out of a job just yet.
What happened: Yesterday, Meta announced four new generations of its Meta Training and Inference Accelerator chips (aka MTIA), built with manufacturing support from Taiwan Semiconductor Manufacturing Company. The chips are strictly for internal use and will “support ranking and recommendations, along with GenAI workloads,” according to Meta. The first version, the MTIA 300 chip, has already been deployed, while the 400, 450, and 500 will follow on a six-month cadence. Unlike the powerful, expensive GPUs made by companies like Nvidia, which can be overkill for some kinds of AI compute, Meta’s MTIA chips are mainly designed for inference—the step where trained models are actually used to generate recommendations, answer prompts, or power AI tools inside apps.
The strategy: Custom chips let companies design hardware around the specific AI tasks they run most often. For Meta, that means optimizing for recommendation systems and AI features inside apps like Instagram and Facebook. Inference is expected to account for about two-thirds of AI computing this year, according to Deloitte, and purpose-built chips can often run those tasks more cheaply and efficiently than general-purpose GPUs.
Everyone else is doing it: Google has built its custom TPUs for over a decade now. Microsoft introduced its first AI chip in 2023, and Amazon has been making its Trainium chips since 2020. Even OpenAI is developing its own custom silicon. Meta first revealed its own chip ambitions in 2023, making it a somewhat late arrival.
The hard part: Obvious statement incoming… but designing competitive chips is notoriously difficult. And just last month, The Information reported that Meta scrapped a more advanced training chip codenamed Olympus after running into several development roadblocks.
Plus, Meta’s timeline is aggressive, trying to crank out chips faster than Marvel puts out movies. “It’s unusual for any silicon company or team to be releasing a new chip every six months,” Meta VP of Engineering Yee Jiun Song acknowledged to CNBC. Even if Meta succeeds, its chips won’t cover all its AI compute needs. Over the past month, it signed: a multibillion-dollar deal for Google’s TPU chips, a $60 billion deal with AMD, and a sweeping new deal with Nvidia. —WK
Patrick’s Take
Meta investing in custom chips makes sense as it prepares for a future in which data centers are expected to shift from mostly training AI models to mostly running them.
Does the move pose a challenge to Nvidia? “To some extent the answer is yes,” says Gaurav Gupta, a VP analyst at research firm Gartner. “But the overall demand is stronger, and the market is growing, so there is plenty of opportunity for all.”
As Meta’s partner on these chips, Broadcom is another winner here. Gupta said the chip giant is “pretty much the partner of choice for all companies going [the custom] route.”
Delivering on these custom silicon ambitions would help Meta stay competitive in the AI race with more energy-efficient infrastructure and a host of chip options that lessen reliance on Nvidia. But it also adds to the massive capital expenditure bills Meta is racking up from AI, which tend to make investors skittish. —PK
THE ZEITBYTE
A hacker’s moral crisis
Morning Brew Design, Getty Images
It’s a case strange enough to be an X-File: A foreign hacker unwittingly broke into an FBI server in 2023, found child exploitation files related to the Epstein investigation, and was so revolted they threatened to report its owner to… the FBI.
According to a Reuters exclusive from yesterday, this breach hit the bureau’s New York Field Office after a special agent inadvertently left a server exposed while navigating the bureau’s labyrinthine digital evidence procedures. After the hacker left a message vowing to turn in the owner, FBI officials talked him down by hopping on a video call and flashing their credentials on camera. The bureau called it “an isolated” incident. In other words: “This is embarrassing; please stop talking about it.” Reuters couldn’t determine the hacker’s identity, country of origin, or whether anyone bothered to pursue them. It’s not clear which specific files were viewed or if they were downloaded. But one Georgia Tech researcher noted that the Epstein files are likely a target for foreign intelligence agencies, telling Reuters, “Who wouldn’t be going after the Epstein files” if they’re “somebody interested in kompromat?”
The FBI fumble is just one example of discomfitingly loose government security practices in recent years. On Tuesday, the Washington Post reported that a whistleblower alleges a former DOGE engineer walked out of the Social Security Administration with a thumb drive containing the sensitive information of 500 million Americans—so check your credit reports. The former DOGE employee apparently told his new coworkers that he expected a presidential pardon if caught. DOGE staffers have also reportedly installed unauthorized servers and a Starlink network at federal agencies. Here’s hoping it won’t happen again—but if it does, at least the next hacker will know who to report it to. —WK
OPEN TABS
- Apple’s foldable iPhone could come with an iPad-style UI—and skip Face ID.
- Tesla wants cars with… no steering wheels. Enter at your own risk.
- Free versions of the top 5 chatbots suggested meal plans “so low in calories and essential nutrients” that they could stunt growth.
- Your Windows 11 PC will soon cosplay as an Xbox—“Xbox mode” is coming next month and could include classic games.
- Pro-Iran hackers claim they hit a major US medical device maker, triggering a “global network disruption.”
- Turns out, the guardrails aren’t… guarding: 8 in 10 AI chatbots are allegedly willing to help plan violent attacks.
- Grammarly’s Expert Review is on ice (for now) thanks to backlash and a class action suit. Better luck next time if you want your LinkedIn posts to sound a little more like Adam Neumann.
- The best way to prove you didn’t write that email using AI? “Typomaxxing.”
This time last week…
Readers’ most-clicked story was about Anthropic CEO Dario Amodei probably wishing his “strongly worded memo” (after OpenAI announced its Pentagon deal) didn’t leak.
Written by Carlin Maine, Whizy Kim, Patrick Kulp, Saira Mueller, and Alex Carr