The Controlled Demolition of the AI Hype Cycle: Why Tech Giants Want to "Freeze" the Bubble
The companies spending hundreds of billions to win the AI race are now asking for permission to slow down together. Dario Amodei’s proposal to “pace the frontier” would bring outside safety reviewers into AI labs and encourage coordination between competitors [1]. He presents it as a response to increasingly dangerous AI capabilities [2]. It also offers the industry something commercially useful: a way to retreat from its spending commitments without admitting those commitments were a mistake.
Economic pressure
The investment depends on a simple expectation: keep adding computing power and data, and the models will improve enough to justify the expense. Diminishing returns threaten that expectation [3]. A model can improve technically while still failing to earn back what it cost to build. If a $10 billion cluster produces only a marginal improvement, investors will eventually question the next $10 billion request.
A slowdown would help control costs, but no lab wants to be the first to cut spending and risk falling behind. It risks alarming investors while giving competitors an opening. A coordinated pause changes that calculation, especially if the industry can explain it as a safety obligation. The labs could reduce their spending while preserving the belief that much more powerful systems remain just beyond reach. A technical plateau becomes easier to sell when the explanation is that progress had to stop because it was too dangerous.
Who sets the rules
That is why the antitrust request in Amodei’s essay matters. He asks Washington for narrow waivers that would let competing labs discuss safety without the usual legal obstacles to coordination. Those conversations could help determine when models are released and what conditions they must meet. The companies with the largest investments would gain a role in setting the pace of the market they dominate.
The public warnings make that request easier to sell. Anthropic researcher Jacob Coxon’s resignation and warnings about AI became international news, with colleagues reinforcing his concerns. A 27 year old engineer leaving a job rarely gets that kind of attention. His media tour gives the industry a compelling story for why governments need to intervene. Whatever his intentions, the story also serves the companies’ commercial interests.
Musk initially dismissed the warnings as a “psyop” [4]. After Amodei published his proposal, Musk declared “Dario is right,” while Altman committed to independent safety reviewers working inside the labs [5]. That reversal does not establish a coordinated deception, but it shows how quickly rivals can agree when a proposal suits their shared interests. They all face the cost of continuing the race, and none wants to slow down alone.
Historical precedent
The 1948 television freeze illustrates what can happen when regulators pause entry into a market to resolve a technical problem [6]. Broadcast interference led the FCC to stop issuing new television licenses while it revised channel allocations [7]. Existing stations continued operating throughout the freeze, which lasted until 1952, while prospective competitors waited. RCA and NBC benefited from that delay regardless of the technical reasons behind it.
DuPont’s support for restrictions on CFCs under the Montreal Protocol shows a related incentive [8]. Its existing chemicals faced declining profitability, while replacements offered new markets and potential patent protection. Supporting restrictions could therefore serve both an environmental purpose and the company’s financial interests [9]. The environmental problem did not have to be invented for DuPont to benefit. The relevant question is what the rules allowed established producers to gain at everyone else’s expense.
Who bears the cost
For AI, that advantage could come through the cost of compliance. Labs backed by Google and Amazon can afford independent safety reviewers and teams of lawyers. A small company or open-source project has fewer resources to meet the same requirements. If governments make those requirements a condition of competing, they risk protecting the largest labs from the pressure that would otherwise force prices down.
That is how the public could end up carrying part of the cost without a direct taxpayer bailout. Developers would pay to satisfy rules shaped by their competitors, while customers could face higher prices and fewer alternatives. Investors would retain some protection for their valuations, and everyone outside the established labs would have less freedom to build around them. Private investments would receive government protection at the expense of people who never made those investment decisions.
The government should not buy the false narrative of a manufactured AI safety crisis. These companies should face the market consequences of a technical plateau. If the bubble bursts, the losses should be theirs to bear, not socialized further.
Links
[1] "We Must Pace the Frontier" Essay — Dario Amodei's original three-step proposal outlining the call for independent safety reviewers working inside the labs and democratic pacing.
[2] The Guardian's AI Slowdown Coverage — An investigative look at Jacob Coxon's viral resignation and how the executive team at Anthropic immediately validated the doom narrative.
[3] Forbes: The Data Plateau and AI Scaling Wall — A detailed analysis on why trillion-dollar infrastructure builds are generating diminishing returns.
[4] The Guardian: Musk dismisses AI warnings as a “psyop” — Coverage of Musk’s response to warnings from Anthropic researchers.
[5] “Dario is right”: Musk and Altman back Anthropic CEO on slowing AI down — Coverage of the executives’ responses to Amodei’s proposal.
[6] Wikipedia: The FCC and Broadcast Interference History — Background on the administrative history of the FCC, frequency constraints, and the 1948 licensing halt.
[7] EBSCO: Historical Channel Allocations — Exploring how the 1948–1952 television pause artificially reshaped the broadcasting landscape.
[8] Penn State University: DuPont's Regulatory Strategy Analysis — Research paper detailing how international environmental regulation offered DuPont a brand new, highly profitable patented market.
[9] The Breakthrough Institute: CFC Alternatives and Innovation Economics — Data tracking how corporate patent strategies map directly onto global climate and safety pacts.