A well-known Silicon Valley podcast, 《All-In Podcast》, discussed the AI industry and made a notably sharp observation: America’s attitude toward AI is shifting negative, and the most concrete outlet for that sentiment is the data centers that AI companies are racing to build across the United States. The source of this backlash could include AI doomsday narratives, fear of unemployment, or—more deeply—a sense of dissatisfaction: this new wave of technological innovation seems to only make a small number of people rich, while most people’s lives show no clear improvement.
U.S. local governments have already overturned data center construction cases
On the show, Chamath Palihapitiya said the problems facing the AI industry right now are not only model competition, capital expenditures, or a shortage of compute power, but that “Americans overall are becoming increasingly averse to AI.” He suggested that the source of this aversion could include AI doomsday theories, fear of job losses, or a deeper form of resentment: this new wave of tech innovation seems to again turn a few into mega-rich—“forging a batch of $1 trillion billionaires,” even—while most people’s lives don’t improve noticeably.
Chamath believes that once this sentiment accumulates to a certain point, the action most likely taken by local communities is to oppose data centers. He gave an example: in the U.S., a local government had originally approved a $6 billion data center construction plan, but then the committee members who supported the proposal were replaced in the subsequent election, and the new officials tried to overturn the original decision. He argued that this shows data centers have become not just infrastructure, but a political symbol for the AI industry and tech billionaires.
Another host, David Friedberg, offered a more direct explanation. He believes many Americans are, in fact, “really starting to hate rich people,” and that data centers have become a physical projection of this feeling. He described data centers as one of the most visible physical spaces for wealth creation in the U.S., and also a machine—through the eyes of ordinary people—by which technological elites, political connections, and billionaires continue to widen the gap.
Friedberg said that for ordinary people, the benefits of AI are not specific enough. Many hear every day that AI will change the world, reshape companies, and increase productivity—but in their own lives, the improvements they truly feel may only be using ChatGPT for medical advice, writing letters, or looking up information. By contrast, what they feel more directly are anxieties about jobs being replaced, concerns that electricity prices may rise, and the fact that tech companies are building massive data centers to train models.
So Friedberg likened data centers to “the mansion tax target of this era.” If politicians in the past attacked rich people’s second homes, mansions, or private jets, then in the AI era, data centers are the new entry point for attacks. They represent progress for tech billionaires, but also represent progress that others do not feel.
David Sacks added from a policy and industry perspective that the reason data centers have become unpopular across multiple U.S. states can be broken down into several categories. First, many local communities worry that data centers consume huge amounts of electricity, which would drive up electricity bills for ordinary households. Sacks said that in some cases, developers in the past really did start seeking local government approvals before they had clear solutions for power, which triggered backlash among local communities.
Second is the combination of AI doomsday groups and the anti–data center movement. Sacks said that some groups arguing that AI could bring catastrophic risks have gradually found it is not easy to directly persuade the public with “AI will lead to endings,” but if they shift the pitch to water and power consumption from data centers, or damage to communities, it becomes easier to mobilize local opposition. He therefore criticized that some anti–data center movements behind the scenes amount to “packaged NIMBYism.”
David Sacks criticized Anthropic AI doomsday narratives
Sacks pointed the finger at Anthropic. He thought that in the past, Anthropic politically allied itself with AI doomsday narratives and NIMBY groups—perhaps because Anthropic didn’t plan to build large-scale data centers on its own, but instead relied on hyperscalers to provide compute power. As a result, opposing data center construction was essentially “throwing sand” on the path of competitors like OpenAI and xAI.
But as Anthropic’s own scale grows and compute demand explodes, if it also eventually has to enter the data center building race itself, that strategy could end up hurting itself.
The episode also mentioned that one of AI companies’ biggest bottlenecks right now is insufficient compute. Chamath pointed out that the stock surge following the market’s reaction to Allbirds’ shift into the AI data center concept looks absurd, but it reflects that capital markets have already realized “compute is extremely scarce.” He said the AI industry not only lacks GPUs, but also lacks land, electricity, data center shells, and local government approvals.
This puts AI companies in a paradoxical situation: on one hand, companies like OpenAI, Anthropic, xAI, and Meta all need more data centers to support model capabilities and revenue growth; on the other hand, public aversion to data centers is getting stronger, and local governments and residents are increasingly likely to block these builds.
Chamath warned that if leading AI companies can’t secure enough compute power, revenue growth might not slow because the products aren’t good enough—it could be because of a problem similar to Friendster’s in the past: demand clearly exists, but the infrastructure can’t keep up, and the company is ultimately overtaken by competitors.
Sacks also believed that if data center construction in the U.S. faces too many restrictions, compute power may move elsewhere—such as regions with cheaper energy, more friendly policies, or even U.S. ally countries. He pointed out that if the U.S. both restricts domestic data centers and opposes allies using U.S. technology to build AI infrastructure, in the end it will only weaken America’s own advantage in the AI race.
Silicon Valley investor: Altman and Amodei aren’t suitable as industry spokespersons
But what is most noteworthy in the episode is the assessment of the AI industry’s PR crisis. Host Jason Calacanis said bluntly that one of the industry’s biggest problems right now is that the people representing the industry speaking out are too bad. He compared America’s perception of AI with China’s highly positive attitude toward AI, arguing that the AI industry in the U.S. currently communicates messages to the outside world that almost all revolve around fear, unemployment, and elite monopolies.
Jason also specifically named that the current public image of the AI industry is related to its representative figures as well. He believed that Anthropic CEO Dario Amodei has long described AI in terms of disasters, cybersecurity risks, and large-scale unemployment, which tends to deepen public fear. Meanwhile, OpenAI CEO Sam Altman, because he has long been at the center of controversy, also can’t take on the role of persuading the public. Jason said directly that these two people “cannot become spokespersons for this industry.”
To improve the public perception of the AI industry, it must be redefined by people who can better explain public benefits like healthcare, education, and housing.
He argued that the AI industry must pull the narrative back toward three directions that can truly improve ordinary people’s lives: healthcare, housing, and education. In other words, AI companies can’t only tell the market how many trillions in valuation they can create, and they can’t only tell enterprise customers how much labor cost they can save—they must help ordinary people see how AI makes medical care cheaper, education more efficient, and housing problems easier to solve.
This article: Sam Altman and Dario Amodei are both too annoying! AI doomsday narratives and relative deprivation have made Americans averse to AI. First appeared on Chain News ABMedia.
