The artificial intelligence boom that has propelled a handful of companies to trillion-dollar valuations may be heading toward a harsh financial reality, according to Yann LeCun, founder of AMI Labs and one of the most influential figures in modern AI research.

In a fresh broadside against Elon Musk and his artificial intelligence venture xAI, LeCun argued that the company is struggling to remain competitive at the cutting edge of AI development and warned that the broader industry faces a reckoning as soaring costs collide with uncertain paths to profitability.

His comments revive a long-running feud between two prominent figures in the AI industry and come at a time when investors are increasingly scrutinizing the economics underpinning the sector’s explosive growth.

“xAI is kind of a failure, frankly, because the founding team has departed,” LeCun said in an interview with CNBC.

Earlier this year, Musk merged xAI with SpaceX in a transaction that valued the combined business at roughly $1.25 trillion, instantly placing it among the most valuable technology enterprises in the world.

Yet LeCun suggested that xAI faces a fundamental challenge that goes beyond technology.

“Elon is now in a position that is very, very difficult for him to kind of hire top people in AI, because he’s kind of, you know, not behaved in sort of very good ways toward the … previous team,” he said.

The criticism follows a series of high-profile departures from xAI over the past year, raising questions about talent retention at a time when competition for elite AI researchers has become one of the industry’s most intense battlegrounds.

Talent Wars Intensify

The battle for AI talent has become as important as the race for computing power. Leading companies, including OpenAI, Anthropic, Google, Meta, Tencent, Alibaba, and DeepSeek have been aggressively competing for a relatively small pool of world-class researchers capable of developing frontier AI systems.

Recent reports from China have highlighted how AI companies are increasingly worried about talent defections. DeepSeek reportedly even required prospective investors in its latest funding round to agree not to poach employees or encourage them to launch competing ventures.

Against that backdrop, LeCun’s suggestion that Musk may be struggling to attract elite researchers strikes at the heart of xAI’s long-term competitiveness. While Musk has invested heavily in computing infrastructure, LeCun argued that infrastructure alone is not enough.

“I’m not very positive about the prospect of xAI,” he said, adding that he does not expect the company to compete effectively with industry leaders OpenAI and Anthropic.

Massive Infrastructure, Mounting Losses

The criticism comes as the economics of AI development face increasing scrutiny. Building frontier AI models requires enormous investments in specialized chips, data centers, and energy. Companies are spending tens of billions of dollars annually to train and operate increasingly sophisticated models.

According to recent financial disclosures, SpaceX’s AI division, which includes xAI, recorded a $2.5 billion operating loss in the three months ended March 31.

LeCun noted that the economics of AI remain deeply challenging.

“The prices are going up of those AI services, but the cost of running them is going down, but not nearly fast enough,” he said.

“And so all of those companies are losing money, and basically, the use for most people is funded by the investors. That can’t go on for a very long right?”

The comments echo growing concerns across the industry that many AI services remain heavily subsidized by venture capital and public market investors. Even OpenAI CEO Sam Altman recently acknowledged that AI costs remain a significant issue, reportedly noting that companies are more focused on the amount they spend on AI services.

LeCun’s most striking warning was directed at the broader AI sector rather than xAI alone. He noted that current business models may prove unsustainable unless companies either raise prices, cut costs, or discover new revenue streams.

“The prices are going up of those AI services, but the cost of running them is going down, but not nearly fast enough,” he said.

“The AMI Labs founder added that labs like OpenAI and Anthropic are ‘going to have to increase prices, they’re going to have to cut costs, or there’s going to be a big bubble explosion.’”

That warning arrives as investors continue pouring money into AI companies at unprecedented valuations. Anthropic recently achieved a valuation approaching $1 trillion. OpenAI has been valued at hundreds of billions of dollars. DeepSeek has reportedly surpassed a $50 billion valuation after its first external funding round.

Meanwhile, infrastructure spending across the sector continues to soar, with major technology companies expected to spend hundreds of billions of dollars this year alone on AI-related investments.

A Deeper Debate About AI’s Future

LeCun’s criticism is not merely financial. It reflects a broader philosophical disagreement over the direction of artificial intelligence.

While companies such as OpenAI, Anthropic, and xAI continue to build increasingly powerful large language models (LLMs), LeCun remains skeptical that current approaches can ultimately deliver truly reliable artificial general intelligence.

Instead, he has championed what he calls “world models.”

Large language models learn patterns in language and predict what comes next, making them highly effective for applications such as coding, writing, and reasoning tasks. World models seek to understand how the real world functions through cause and effect, physical interactions, and environmental awareness.

“I personally don’t think we’re going to have generalized reliable agentic systems until they’re based on world models,” LeCun said.

Much of the AI industry is betting heavily on autonomous AI agents capable of performing complex tasks with minimal human supervision. Companies from OpenAI and Anthropic to Qualcomm and Salesforce are investing heavily in agent-based systems that they believe will transform software, customer service, and productivity applications.

LeCun, however, believes current architectures may ultimately hit limits.

The debate centers on whether the current AI boom can generate sustainable profits. Many of the industry’s leading models require enormous computing resources to operate, creating high ongoing costs even after development is complete.

LeCun suggested that this imbalance between operating costs and customer willingness to pay remains unresolved.

“LLMs are useful for areas such as coding or math,” he acknowledged.

But, he added, “the cost of running those systems with this kind of performance is very high compared to the amount of money that users are ready to pay.”

That challenge has become more visible as competition intensifies. OpenAI is reportedly considering significant pricing changes. Anthropic has faced growing pressure over model costs. Hardware providers are spending heavily to support the AI ecosystem, while investors increasingly demand clearer paths to profitability.

Morale inside Meta has plunged to what insiders describe as some of the lowest levels in the company’s history, as the social media giant’s layoff, tied to an aggressive push into artificial intelligence, exacts a heavy human cost on its remaining workforce.

The frustration boiled over last month when Meta laid off approximately 8,000 employees, roughly 10% of its total headcount, in yet another round of sweeping cuts tied to the company’s frantic refocusing on AI. Those who kept their jobs now find themselves increasingly saddled with repetitive, low-level tasks to train and refine AI models, a grind that many say is draining what little enthusiasm remains.

According to Wired, in an internal memo sent to staff on Friday, CEO Mark Zuckerberg tried to rally the troops by announcing a companywide AI hackathon scheduled for July. The response was swift and brutal, exposing a deep disconnect between leadership’s vision and the day-to-day reality for employees still reeling from cuts.

“I’m literally preoccupied with keeping the lights on for my team,” one employee wrote in an internal message quoted by Wired. “I have no incentive to participate, let alone have the time to do so.”

Another added, “I’m not sure that this company supports a hackathon culture anymore. People are being asked to cover more work with less support while their colleagues get laid off.”

A third worker noted: “I’ve participated in previous hackathons, but this no longer feels like an option alongside pod sprints in my corner of the company.”

Zuckerberg’s additional gesture, offering employees access to permanent desks instead of the controversial “hot desking” system, only seemed to underscore how precarious many roles had become. The move, intended as a positive signal, landed as tone-deaf amid widespread anxiety about job security.

A Painful Transition with Limited Payoff So Far

Meta’s difficulties stem from the broader challenges facing Big Tech as it races to catch up in the generative AI era. Despite heavy investment, the company continues to struggle to produce standout models that match the pace set by rivals like OpenAI, Anthropic, and Google. Insiders say the pressure to deliver results is intensifying, but the immediate human cost is mounting faster than visible breakthroughs.

Andrew “Boz” Bosworth, Meta’s chief technology officer, was unusually candid during an internal “Tuesdays with Boz” meeting on June 2. According to four people familiar with the call, Bosworth described the current atmosphere as “maybe not the worst it’s ever been in 20 years here, but it’s probably up there” — and “probably one of the worst it’s ever been.”

The only period he recalled as worse was the 2018 Cambridge Analytica scandal, when revelations about the misuse of millions of Facebook users’ data for political targeting triggered a massive crisis of trust, regulatory scrutiny, and reputational damage. At the time, whistleblower Christopher Wylie exposed how the firm had harvested data to influence campaigns, including Donald Trump’s 2016 presidential run and the UK’s Brexit referendum.

Zuckerberg himself acknowledged the difficulties in his memo, admitting that the AI transition has been messy.

“Given the complexity of these changes, we’ve made mistakes and will almost certainly make more,” he was quoted as saying.

He pledged to avoid further layoffs for the rest of the year, but the damage to trust appears significant. Employees who survived previous rounds now face heavier workloads with fewer resources, fueling resentment toward a leadership team that continues to emphasize long-term AI bets over near-term stability.

The Big Tech’s AI Reckoning

Meta’s situation is not unique, but its scale and visibility make it a bellwether for the industry. Across Silicon Valley, companies are pouring tens of billions into AI infrastructure while simultaneously trimming headcount to improve efficiency. The result is a workforce that feels both essential to the future and expendable in the present.

For Meta specifically, the pivot carries extra weight. The company’s core advertising business still generates enormous cash flow, but growth has slowed as users fragment across platforms and regulators tighten oversight. AI is seen as the next growth engine, powering everything from content recommendation to ad targeting and new products, yet translating that vision into reality has proven slower and more painful than expected.

The hackathon misstep highlights a deeper cultural challenge. What once felt like an innovative, energizing part of Meta’s identity now strikes many as tone-deaf busywork when teams are stretched thin, and colleagues have just been shown the door.

Industry observers warn that sustained low morale could undermine Meta’s ability to attract and retain top talent at a time when competition for AI experts is fierce. If the best engineers and product minds start looking elsewhere, the company’s ambitious roadmap could slip further behind.

Zuckerberg’s memo and Bosworth’s comments are seen as a suggestion that leadership recognizes the gravity of the situation, even if their proposed remedies have so far fallen flat. As Meta bets its future on artificial intelligence, it is painfully discovering that keeping its human workforce engaged may be one of the hardest problems of all.