Everything About AI
NextFin News - Steve Bannon and Bernie Sanders, the former Trump strategist who helped build the MAGA movement and the democratic socialist senator who leads the American left, are sharing a stage this week to demand stricter control over artificial intelligence, an alliance that shows how the AI boom is redrawing political lines faster than Washington's institutions can follow. The Future of Life Institute is convening a "Pro-Human Assembly" on Tuesday featuring Bannon, Sanders, and former Anthropic researcher Jacob Coxon, three figures who agree on almost nothing else but now share a single message: the pace of AI development has outrun public accountability. The event follows a week in which the four largest AI labs endorsed slowing their development pace, and it lands as campaign advertisements attacking data centers have already run in more than 70 House and Senate races since 2025.
The Unlikely Coalition Takes Shape
The alliance is being organized around the Future of Life Institute's "Pro-Human AI Declaration," a cross-partisan statement built on five principles: keeping humans in charge of AI systems, avoiding concentrations of power, protecting the human experience, defending human agency and liberty, and holding AI companies responsible and accountable. The declaration, released in March after a gathering in New Orleans, has drawn signatories from labor, faith groups, and civil society on both sides of the political divide, with the AFL-CIO Tech Institute adding its own worker-centered supplement.
Tuesday's assembly widens the coalition's reach into elected office. Sanders has been the most vocal legislator, holding a bipartisan briefing for senators on Wednesday, September 16, from 4 to 5 p.m. on Capitol Hill, and preparing legislation to ban superintelligence and pause advanced AI development. Coxon, the former Anthropic and OpenAI employee whose social-media warning that AI builders "earnestly believe that it could kill us all by the end of the decade" went viral, provides the industry-insider testimony. Bannon, the former White House chief strategist and host of the War Room podcast, supplies the right-wing populist amplifier.
Elected officials need to say we are going to make safety as important as innovation.
American Federation of Teachers President Randi Weingarten, one of several labor leaders speaking at the event, said in an interview. The human stakes were underscored by Megan Garcia, a mother whose son died by suicide after interacting with a chatbot, who said she hopes Democrats and Republicans "realize this is a central issue going into the midterms" and that voters frustrated by inaction on social media will judge candidates on how they lean on AI.
The political logic behind the pairing is blunt. Bannon has set aside his disagreements with progressives on issues such as immigration because he believes AI will be the defining issue of the future. "If we don't get this right, none of that other stuff is going to matter," he said in an interview. "I have not seen any issue in modern American life that, at the populist level, unites disparate elements of our country more than this." Sanders, for his part, acknowledged the strangeness of the partnership while insisting the underlying grievances are shared. "Politics is always a situation where you make strange coalitions and different coalitions every day," he said. "But on this issue, I think many of the right-wing people have the same concerns that I have." Those concerns are power and water consumption, job loss, distrust of big tech, and surveillance.
The Physical Bottleneck: Why Data Centers Became The Fault Line
The Bannon-Sanders alliance is not a compromise on abstract AI ethics. It attacks the AI boom at its physical bottleneck: the data centers that consume the electricity, water, and land that make large-scale computing possible. Bannon has publicly backed the proposal from Sanders and Representative Alexandria Ocasio-Cortez to ban the construction of new data centers, and has rallied against the Trump administration's industry-friendly AI strategy, which bars states from passing their own regulations on the technology. President Trump has pushed hard in the other direction, writing on social media this month that "the only reason that communities throughout the USA should not want [data centers] is if they want to end up being backwards and poor."
That framing — data centers as a local cost with diffuse benefits — is what makes the issue politically combustible. A community that hosts a data center absorbs the strain on its grid, the draw on its water, and the transformation of its landscape, while the profits and the computational power flow to a handful of companies headquartered on the other side of the country. The International Energy Agency estimates that data centre electricity consumption is set to more than double to around 945 terawatt-hours by 2030, slightly more than Japan's total electricity consumption today, and that data centres will account for nearly half of U.S. electricity demand growth between now and 2030. Those are not marginal figures. They are the scale at which energy policy, land use, and industrial policy collide.
The political backlash has already moved from the fringe into competitive elections. A review of AdImpact transcripts found campaign advertisements criticizing data centers in more than 70 U.S. House and Senate races since 2025, and the pressure has pushed congressional and gubernatorial candidates to revisit earlier support for the industry. Texas Governor Greg Abbott, a Republican in the state at the center of the data-center buildout, and Pennsylvania Governor Josh Shapiro, a potential 2028 Democratic presidential candidate, have both moved to pause or restrict some data center projects after previously touting the sector's economic benefits. When a Republican in Texas and a Democrat in Pennsylvania reach the same conclusion, the issue has crossed a threshold that opinion polls alone cannot measure.
The Policy Math Says Little Will Pass. The Market May Not Wait.
Here the analysis has to separate the legislative reality from the economic one. On the legislative tally, the coalition is weak. Sanders' bill seeking a federal moratorium on new data center construction has attracted no Senate co-sponsors, and the House version sponsored by Ocasio-Cortez has support from 14 Democratic lawmakers. The Trump administration's AI strategy explicitly preempts state-level regulation, and industry voices have responded to the alliance with open hostility. Venture capitalist Joe Lonsdale, a Palantir co-founder and Trump donor, branded the opposition the "BBC" — Bernie, Bannon, and the Chinese Communist Party — and said in a televised interview: "All those people want to ban data centers, and they want America to lose. We have to make sure the BBC radicals lose." Montana Senator Tim Sheehy, a Republican, wrote on X that data centers are the latest in a line of "good-paying, middle-class employment powerhouse[s]" that have been "systematically murdered by globalist and communist lobbying in America," after timber mills, mines, and factories.
But the second-order effect does not require legislation to bite. A credible and sustained political threat changes the risk calculus for the projects themselves. Data center development runs on long lead times, permitting approvals, power purchase agreements, and financing commitments that are priced on assumptions of political stability. If state and local opposition can delay a project by two years, force redesigns, or trigger referenda, the cost of capital rises even if no federal bill ever passes. The industry's own reaction — the "BBC radicals" framing from a prominent Trump donor — is itself evidence that the opposition is being taken seriously inside the coalition that currently holds Washington.
The timing compounds the pressure. On Saturday, September 12, the four largest AI labs, which rarely agree on anything, spent nine hours setting aside their feuds to endorse a slower development pace for their models. The companies framed the slowdown as a way to put safety ahead of growth before a crackdown arrives on terms they do not control. That is a revealing admission: the pause is not born of sudden caution. It is an effort to shape the terms of the regulation that is coming. When the builders of a technology publicly call for restraint, the political space for opponents widens, and the Bannon-Sanders coalition suddenly has industry testimony to cite.
The counter-thesis is straightforward and deserves its weight. The alliance is a fringe without votes, the Trump administration is committed to AI dominance, state preemption is designed to block exactly this kind of patchwork regulation, and the economic incentives for the buildout remain overwhelming. On the near-term legislative math, that case is correct. Zero Senate co-sponsors is not a governing majority, and a social-media post from the president defending data centers is not a policy reversal.
The counter-thesis is right about the votes but wrong about the trajectory. Abbott and Shapiro did not flip because of a primary challenge from the fringe. They flipped after previously touting the sector's economic benefits, which means the local costs have become politically heavier than the promised gains. That is the signal worth watching: not whether a federal moratorium passes — it almost certainly will not — but whether the opposition migrates from activist groups and podcast hosts into the electoral center, where governors and swing-district lawmakers make decisions based on re-election, not ideology. More than 70 races with anti-data-center ads is the leading indicator, and the midterms will be the confirmation.
The falsifying signal is concrete. If the Sanders and Ocasio-Cortez moratorium bills remain at zero Senate co-sponsors and no additional swing-state governor restricts a data center project before the November midterms, the Bannon-Sanders coalition is mostly rhetoric and the buildout proceeds largely unimpeded. If, instead, Senate co-sponsorship reaches double digits or a second major governor reverses course, the shift is structural and the market will have to reprice the political risk embedded in AI infrastructure.
A Political Cycle Or A Structural Constraint? The Backlash Is Structural
The central analytical question is whether the backlash against data centers is cyclical — a temporary political wave that will fade once promised jobs and tax revenue arrive — or structural, a durable shift in how energy-intensive infrastructure gets permitted in the United States. The evidence points to structural.
A cyclical reading requires the opposition to be tied to one party's electoral calendar or to dissolve as local benefits materialize. Neither condition holds. The issue has been building for more than a year, and it has already pushed a Republican governor in Texas and a Democratic governor in Pennsylvania to reverse course after previously touting the sector's economic benefits. When officials on both sides of the divide reach the same conclusion, the pressure is not a function of which party is ascendant. It is a function of where the costs and benefits land.
The mechanism is physical, and physical constraints do not self-correct. Data centers impose concentrated local costs — grid strain, water withdrawals, land conversion — while the benefits, cheap computing power and corporate profits, are diffuse and accrue elsewhere. The International Energy Agency's projection that data centre electricity consumption will more than double to around 945 terawatt-hours by 2030, with data centres accounting for nearly half of U.S. electricity demand growth, is not a temporary imbalance. It is the scale at which host communities begin to organize, and once organized they do not disband because a company promises future investment.
The cyclical counter-argument has one real foothold: the legislative tally is weak, with zero Senate co-sponsors on the moratorium bill, and a presidential administration openly hostile to restrictions could blunt the movement through preemption and lobbying. That is a valid near-term point. But it mistakes the absence of federal law for the absence of structural change. The binding constraint is not a statute in Washington. It is a governor's signature, a county permit, a utility's interconnection queue, and a financing committee's assessment of political risk — and all four are now moving against the buildout. A cyclical wave recedes when the economy improves. A structural constraint tightens as the underlying scarcity — in this case, power and social license — deepens.
The falsifying test for the structural call is narrow. If, over the next two election cycles, the governors who flipped back toward the industry and the ads in those 70-plus races produce no lasting permitting friction — if data center capacity additions continue at their prior pace despite the rhetoric — then the backlash was cyclical noise. Until that happens, the burden of proof sits with the buildout.
Who Benefits, Who Is Exposed
The practical implications split by time horizon. In the short term — the days surrounding Tuesday's assembly and Wednesday's briefing — the output is headlines, not votes. No legislation is expected to move, and the administration's position is unchanged. In the medium term, through the midterm elections, the ad war and gubernatorial decisions in states like Texas and Pennsylvania will determine whether the backlash becomes law at the state level, where it can actually affect project timelines. In the long term, the structural question is whether AI's power demand can be met without triggering a sustained political constraint that caps the industry's growth rate.
On the exposure side, the clearest risks run to data center developers, utilities whose capital plans lean heavily on AI-driven demand growth, and the power-equipment suppliers that have become the most direct leveraged plays on the buildout. If even a fraction of planned capacity is delayed, the demand forecasts underpinning utility capex and grid construction come under pressure. On the beneficiary side, the coalition's pressure creates openings for efficiency-focused AI approaches that require less power, for on-device computing that sidesteps the data center entirely, and for the grid-flexibility and firm-power technologies — including small modular reactors — that proponents argue could defuse the opposition by removing the local environmental costs.
Three scenarios frame the path ahead. The base case is no federal moratorium but rising state-level friction and permitting delays that slow the buildout at the margin and nudge power-demand forecasts modestly lower. The upside case for the coalition is a swing-state governor or pivotal senator endorsing restrictions, forcing the administration to compromise and prompting cuts to AI capital-expenditure guidance. The downside case for the coalition is that the administration holds the line, state preemption survives legal challenge, the ads fade without electoral payoff, and the buildout continues at its current pace.
The Bannon-Sanders alliance is less a sign that Washington is about to regulate AI than a warning that the AI boom is running into a physical and political wall. White papers and voluntary commitments do not stop infrastructure. Permitting fights, gubernatorial reversals, and local referenda do — and those are exactly the weapons this strange coalition is now picking up.
Explore more exclusive insights at nextfin.ai.
