Public resistance to AI has increased as data center construction has accelerated causing electricity, memory chips, and water prices to rise. One survey found that 70% of Americans are opposed to data center construction in their local area.

Mark Cuban was the first to break with the other tech bros saying this about data centers: “The big LLMs have lost the PR battle. Why? Because they all suck at putting people first.” He accused them of having a Silicon Valley ‘attitude that “makes them all think they are John Galt,” a fictional character who was purportedly saving the world from communism. Now the tech bros are saving us from the Chinese and a worse standard of living.

Some tech bros are even questioning the corporate benefits from generative AI, a year after academic analyses started questioning them. A 2025 MIT Study found that 95% of AI initiatives fail to achieve a positive ROI. A subsequent study from software company Atlassian reported similar levels of failure.

Some claim the problem is AI slop. An article published in Harvard Business Review in 2025 reported that 41% of survey respondents said they had received slop in the previous month and that 15% of the overall content they received was workslop.

This year, Harvard Business Review went further saying that when slop occurs in sequence across a business’s processes, those processes and their outputs start to deteriorate, errors compound, trust erodes, and any productivity gains disappear.

Recent price increases by AI Labs such as OpenAI and Anthropic have contributed to corporate resistance. They had to increase prices because they were so far from profitability amidst plans to do IPOs. Thus, tokenmaxxing, an act of encouraging workers to spend as many tokens as possible to complete tasks is over; token optimization is in. “A UBS Group survey shows that approximately 60% of companies have already implemented guardrails for token usage.” More specifically, “companies are responding with “model routing, directing simpler tasks to lower-cost or even Chinese open-source models.”

What to make of these survey results? We knew that tokenmaxxing and leaderboards were a bad strategy: “one company exhausted a significant portion of its annual token budget and had to reduce its internal AI tools from five to two; another saw a single user incur $35,000 in monthly costs on AWS Bedrock; and some DevOps team members consistently hit 100% to 200% of their weekly token quotas, though their employers have not yet intervened explicitly.”

Nvidia and Palantir

Nvidia vice president Bryan Catanzaro was one of the first tech bro users to react to the rising token prices because of the increases. Catanzaro said that for his team, AI compute now costs more than the employees using it, making AI more expensive than human labor. And he said this despite Nvidia, his employer, benefitted greatly from tokenmaxxing.

Alex Karp, the CEO of Palantir, focused more on the combination of AI slop, data ownership, tokenmaxxing and thus the value of generative AI to companies. “Where do traditional businesses fit in an AI world? Who captures the value created by AI: the companies deploying it or the labs developing the actual AI? More specifically he says: “At every single enterprise I deal with, these people are livid, they’re like, I am paying for tokens that create no value.”

Microsoft

Microsoft CEO Satya Nadella is beginning to say similar things. He sees an imbalance between businesses and AI labs of which Microsoft has been a big investor in one, OpenAI. He is concerned with the soaring costs and also who retains the data.

He has been telling other corporate chieftains about his concerns that companies should be able to retain the learnings that come from using AI models. “If you’re just a consumer of a foundation model, then I’m not sure how you can retain enterprise value, let alone create.” One aspect of this is the “concerns over the power and insight AI labs derive from their business customers’ data and decision-making, the magic sauce behind what makes a company successful.”

What is Nadella’s solution? “We are making a $2.5B investment in Microsoft Frontier Company, embedding 6,000 industry and engineering experts at customers to co-design, co-innovate, deploy and continuously improve AI systems at a scale based on measurable business outcomes.”
A journalist says: “I don’t know what most of that means, and I’m not entirely sure [Microsoft] does either, but what I can tell you is that Amazon, Anthropic, and OpenAI all launched similar programs recently, with the idea being that simply having companies subscribe to LLMs and spend outrageous amounts of money on monthly token allotments has proven financially ruinous for everyone involved.”

Zuckerberg and Palihapitiya

Zuckerberg has also chimed in with his criticisms: the “trajectory of the agentic development over ⁠at least the last four months hasn’t really accelerated in the way that we expected, and that the company’s bets on the new structure haven’t come to fruition yet. Zuckerberg was referring to AI agents, automated systems that can execute tasks on behalf of a user, for which his top people were very optimistic in January and February.

Finally, Chamath Palihapitiya, an early Facebook executive who now runs Social Capital, and which raised big money for SPACs a few years ago. “The AI labs are manufacturing panic about their own technology to raise money. They sell the same technology as a miracle to raise money, then sell it as a weapon to bring in the regulators.” As one commentator said: “When they need money, the message is that they have created a super god. The venture capitalists then behave like lemmings, because nobody wants to be caught on the wrong side of a super god.”

The one executive who is still super positive is Masayoshi Son, CEO of Softbank, which is the biggest investor in OpenAI. He “groused that any talk of an AI bubble is absurd, adding that he’s willing to waste trillions of dollars a year for decades if that’s what it takes to make AI a financial success.”

Last year, it was almost unheard for the tech bros to criticize generative AI. Instead, they were singing the praises of AI like Masayoshi Son still does, claiming it was going to be the biggest technology ever. Now, cracks are appearing in their facade.