How China's DeepSeek may have made Mark Zuckerberg lose faith in Meta's AI team, and start the biggest-ever talent war in Silicon Valley
Zuckerberg
is preparing to defend billions in AI spending as
Meta
faces its slowest profit growth in two years, with the CEO creating a secretive "
Superintelligence lab
" after the company's latest AI model reportedly failed to meet expectations.
Meta's
Llama 4 AI model
attempted to copy techniques from Chinese startup DeepSeek but received such poor reception from developers, making Zuckerberg authorize a complete overhaul of the company's AI strategy, according to sources cited by CNBC and the Financial Times.
Meta invested $14.3 billion in Scale AI, bringing CEO Alexandr Wang aboard as chief AI officer alongside former GitHub CEO Nat Friedman, ex-Safe Superintelligence CEO Daniel Gross, and ChatGPT co-creator
Shengjia Zhao
to lead the new Meta Superintelligence Labs. The unit operates in isolation at Meta's Menlo Park headquarters, with Zuckerberg personally meeting new hires as they join the closely guarded project.
The catalyst for this dramatic pivot came in January when DeepSeek's R1 model caught Meta off guard. The Chinese lab's mixture-of-experts (MoE) approach appeared cheaper to train and run than traditional dense models, prompting Meta executives to believe they had found a shortcut to leapfrog rivals like OpenAI, sources told CNBC.
by Taboola
by Taboola
Sponsored Links
Sponsored Links
Promoted Links
Promoted Links
You May Like
Health: the best shoes to walk and stand all day
Ultra-Comfortable Shoes
Undo
However, Meta's rush to adopt MoE architecture backfired spectacularly. Developers found Llama 4 more difficult to customize and integrate compared to its predecessor, with many preferring the older Llama 3 model. The disappointing reception led Zuckerberg to lose confidence in his generative AI team's leadership, particularly after controversies over whether Meta had manipulated industry benchmark tests.
Zuckerberg's Superintelligence Lab is Meta's Manhattan Project-equivalent
The new Superintelligence Lab represents Mark Zuckerberg's attempt to create a startup-like environment within Meta's $1.8 trillion structure. "It's like the Manhattan Project. They are throwing all their cash at AI and trying to work out what to do," Uday Cheruvu, portfolio manager at Harding Loevner, told the Financial Times.
The hiring blitz began as early as March, when Zuckerberg started drawing up a list of top AI researchers before embarking on his offensive. He has since wielded nine-figure sign-on bonuses to poach approximately 50 top AI researchers from competitors including OpenAI, xAI, Tesla, Apple, and Google, including some of the biggest names in the industry.
OpenAI CEO Sam Altman said Meta offered some of his employees as much as $100 million, and Wired's reports also indicate the same that Zuckerberg is offering $100 million for the first year, with over $300 million over four years. Although Altman also claimed he was sure no one would leave, since they were dedicated to OpenAI's mission, he turned out to be wrong. Last week, Meta hired ChatGPT co-creator Shengjia Zhao as the lab's chief AI scientist, marking a significant coup in Silicon Valley's intensifying talent war.
Despite the extraordinary compensation packages, gains have been limited: only a fraction of high-profile researchers opted to join. Meta also explored acquiring two startups founded by former OpenAI executives, Safe Superintelligence and Thinking Machines Lab, but was rebuffed in both cases. Then, acording to Wired, Meta approached more than a dozen staffers at Mira Murati's AI startup to discuss joining the lab. One researcher reportedly received an offer of more than $1 billion, but none accepted the offer, at least as of this writing.
Zuckerberg's spending spree hasn't been an isolated event. Microsoft recently poached two dozen Google DeepMind AI researchers, including some working on the company's Gemini model. Meanwhile, Google, after failing to close a deal with another startup on OpenAI's radar, invested $2.4 billion into WindSurf, acquiring its CEO, key AI talent, and its core tech, just as Meta did with Scale AI.
While some believe this AI talent war benefits researchers, DeepMind CEO Demis Hassabis called it a rational move for Meta, saying the company is merely trying to catch up. Others, like Sam Altman, are less charitable, he has compared Meta's tactics to those of the mafia.
Zuckerberg's ambitions go beyond catching up. His goal is Superintelligence: AI that surpasses human intelligence across all domains. This includes integrating advanced AI into Meta's social platforms to offer digital companionship, generate unlimited personalized content, and enhance targeted advertising.
The CEO goal believes assembling an elite team is essential to maintain Meta's position as a top social and advertising platform, especially as rivals like Elon Musk's X are absorbed into AI-first companies. 'You actually kind of want the smallest group of people who can fit the whole thing in their head. So there's just an absolute premium for the best and most talented people,' he told The Information.
Meta faces investor scrutiny over massive AI spending
When Meta reports second-quarter earnings Wednesday, analysts expect the company's slowest revenue growth in seven quarters at 14.7% to $44.80 billion, with profit growth slowing to 11.5% as operating costs jump nearly 9%, according to Reuters data from LSEG. BNP Paribas estimates the Superintelligence hiring spree could drive $1.5 billion to $3.5 billion in additional annual research spending, with Meta already raising its capital expenditure forecast by 10% to between $64 billion and $72 billion for 2025.
Despite the massive investments and internal tensions as existing AI staff worry about being sidelined, Meta's stock has risen 20% this year as investors show more patience for AI spending compared to the metaverse push that drew sharp criticism in 2022. However, executives are now questioning Meta's open-source AI strategy, with some considering abandoning the planned release of Llama 4's "Behemoth" version in favour of developing proprietary models.
Zuckerberg remains confident the bet will pay off, telling The Information that Meta's "very strong business model that throws off a lot of capital" can support hundreds of billions in infrastructure and talent investments. The challenge now is convincing Wall Street that this secretive Manhattan Project will deliver the AI breakthrough Meta desperately needs.
AI Masterclass for Students. Upskill Young Ones Today!– Join Now
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Hans India
an hour ago
- Hans India
Ageing population, high debt seen as drags on China's growth ahead
China is expected to face an adverse economic impact in the coming decades due to its ageing population and high government debt, according to reports. High government debt raises interest costs and leaves less fiscal room to respond to shocks, just as ageing populations push up pension and health outlays, according to a report in Newsweek. The Chinese and US governments are among the most indebted in the world. The US government's gross debt at 123 per cent is equal to the country's GDP, according to International Monetary Fund data. China's stands at 84 per cent, buoyed by debt-driven growth in the 2010s and a housing market crunch that has heavily indebted local governments. London-based global advisory firm Oxford Economics estimates the Chinese economy's potential growth could be cut roughly in half by the 2050s. According to the Newsweek report: "Soaring pension and healthcare expenses are the biggest policy challenge of the 2020s in all advanced economies and most emerging ones." As per a United Nations report, China currently has a median age of around 40, which is well above the global average, and is projected to reach 52 by 2050. This would be much higher than even the US median age, which is expected to stay around 41 years. China's old-age dependency ratio, or the share of people aged 65 and older, is projected to rise by more than 50 percentage points by 2026 compared to 2010, versus roughly 8-10 points in the United States. This will strain China's modest safety net. And unless the country is able to reverse its flagging birth rate, this will shift the burden onto a smaller pool of workers, according to the report in Newsweek. Jed Cartledge, an economist and one of the authors of the Oxford Economics report, said this better positions the US demographically. China's fertility rate of 1.2 births expected per woman is among the world's lowest. While higher, the US rate of 1.6 births remains well below the rate of 2.1 necessary to sustain a population naturally. Cartledge pointed out, however, that historically, immigration has largely offset declining births and averted demographic problems in the U.S. "Admittedly, US immigration is taking a hit under the second Trump presidency, but we're expecting the reduction in net immigration to only last through the remainder of his second term before reverting to a 1.1 million per annum, which was the typical pace prior to the pandemic," Cartledge told Newsweek.
Mint
an hour ago
- Mint
Gamblers now bet on AI models like racehorses
Now that AI developers are getting paid like pro athletes, it's fitting that fans are placing big bets on how well they're doing their jobs. On Kalshi, Polymarket and other sites where people wager 'predictions" on real-world events, gamblers lay down millions each month on their picks for AI's top model. The AI arms race is playing out in plain sight on social media, ranking sites and obscure corners of the internet where enthusiasts hunt for clues. The constant buzz makes the topic appealing for wagers, though not every scrap of information is meaningful. Foster McCoy made $10,000 in a few hours in early August by betting against the success of OpenAI's GPT-5 release. The 27-year-old day trader noticed people were misreading an online ranking site that appeared to hype GPT-5, so he put $4,500 on its competitor, Google's Gemini, to be 'Best AI This Month." As more bettors fell in line with his call, he cashed out. That's just another day for McCoy, who has traded $3.2 million on Kalshi since the start of 2025—making $170,000. He's part of a growing contingent of bettors making hundreds of trades a week on AI markets, on a range of wagers such as 'Best AI at the end of 2025," 'AI regulation becomes federal law this year," and 'Will [Chief Executive] Sam Altman be granted an equity stake in OpenAI this year?" Trading volume across AI prediction markets has surged to around $20 million this month. Kalshi, the only platform currently available in the U.S., is seeing 10 times the volume on AI trades compared with the start of the year, a spokesman says. Each bet, or 'contract," is priced in cents to reflect the odds: McCoy bought thousands of Gemini contracts at around 40 cents, meaning it had a 40% chance of winning. If the bet had settled and Gemini won, McCoy's 40 cents would become a dollar. If Gemini lost, McCoy would lose it all. But much of the action happens before the final outcome. As more people piled into the Gemini bet, the contract price rose. McCoy sold when it had reached 87 cents. It's like betting on a sports match, only with the option to cash out when the odds rise in favor of your bet. 'You're just betting against what the other guy knows," says McCoy, who credits his success to 'being chronically online." He trades from home on a triple-monitor setup tuned to the holy trinity of AI betting: X, Discord and LMArena, a leaderboard where people rate AI model performance in blind tests. Social-media beefs can be gold for gamblers. After GPT-5's debut, a flurry of Elon Musk posts claiming the superiority of his xAI Grok chatbot sent the 'Grok to Win" market up more than 500% within hours. Before long, it had fallen nearly all the way back down. Bettors end up down rabbit holes hunting for more obscure tidbits. Harvard undergraduate Rishab Jain often scans X accounts of lesser-known researchers. The day before GPT-5's release, OpenAI's Sam Altman posted an image of the Death Star from 'Star Wars." When a researcher from Google's DeepMind replied with an image of the killer globe under attack, Jain read it as a sign of confidence from Gemini's makers. Jain also scrapes source files from Google's apps and monitors public GitHub repositories tied to its products, looking for back-end changes that might signal a soon-to-be-released Gemini model. 'I'm almost obsessively up-to-date with what's going on in this world," Jain says. 'Google has so many products that need to integrate a new model before it officially launches, and you can see those changes happening in the back end if you know where to look." Since starting in June, he has won $3,500. Strategies vary. Some bet on the big industry players, others buy low on less-known or soon-to-be-updated models. Some compare odds on Kalshi and Polymarket to find arbitrage opportunities in the odds. As volume for these AI trades continues to grow, the incentive for good information will only increase, and the squeeze on casual bettors will get tighter, says Robin Hanson, a professor of economics at George Mason University. 'When you have better information in these kinds of markets, you can make better decisions," Hanson says. 'If you know a little more, you make more money." For now, the markets still draw a mix of sharks like McCoy and Jain along with casual bettors. 'I'm just a dude that likes tech and has some time and money," says James Cole, a 35-year-old who recently shut down a company he founded. 'I'm speculating with mostly instinct and 10 minutes of research." He says he's up for the year…so far. Write to Ben Raab at
India.com
2 hours ago
- India.com
China's BIG plan to conquer the Moon, to build lunar city, settlements by 2035, develops tech to transform moondust into...
Representational Image/AI-generated China Moon Mission: China's rapid technological advancement has dazzled the world over the last decade, and now the Asian giant has reportedly developed a technology that transforms moondust (regolith) into sturdy bricks which would be used in building houses on the moon, as Beijing plans to establish lunar settlements by 2035. What is China's 'moon bricks' technology? According to media reports, Chinese scientists have developed an advanced machine to build bricks from regolith on the surface of the moon. The machine collects moondust and superheats it a special chambers, where the high-temperatures fuse the dust particles together to form solid bricks that can be used to construct research bases and housing facilities on the moon. The technology is a major step towards establishing settlements in space as it nullifies the cost of transport construction material from Earth, as transportation for a single kilo could run into crores. How the technology works? As per experts, the machine makes bricks similar to how regular bricks are made back on Earth, by baking clay in a hot furnace. The machine heats moondust to a temperature of up to 1300 degrees Celsius by amplifying the available sunlight by 3000 times. The superheated furnace melts the dust and shapes it into durable space bricks. The machine was first tested with volcanic ash on Earth, but melting regolith is a challenging task as its composition varies in different spots on the lunar surface due to radiation exposure, temperature fluctuations and micrometeorites striking the moon. Where will the bricks be tested? China's Tianzhou-8 mission ferried a cache of regolith bricks to space where they will undergo rigorous testing for radiation and temperature effects for three years at the Tiangong Space Station. These tests are designed to determine their durability and efficacy for building installations on the moon. In 2020, China's Chang-e-5 mission brought back soil samples from the moon, and soon Chinese engineers designed a machine to create moondust bricks after studying the composition of regolith. The Chang-e-8 mission, which aims to lay the foundation of a Chinese research base on the moon, will carry the technology with it to the lunar surface in 2028. When China plans to build a lunar city? According to reports, China's Chang-E-7 mission will head to the moon in 2026 to search for water on the lunar south pole, while a manned mission is planned for 2030, and a permanent research base along with a lunar colony, is expected to be operational by 2035. Notably, China and Russia have teamed up to build a nuclear power plant on the Moon, which will power the proposed International Lunar Research Station (ILRS). The revolutionary project is expected to be completed by 2036.
