A fresh wave of AI‑doom headlines is dominating front pages again, and markets are responding in ways consistent with prior cycles: chip stocks are declining, hyperscalers are rising, and investors are evaluating how a potential slowdown in frontier model development could affect different parts of the AI stack.

Jacob Coxon — a researcher who specializes in training new AI models by having them consume massive amounts of data — resigned from Anthropic because he doesn’t want to help build AI systems that can “improve themselves, spiral out of control and destroy humanity.”

This is not the first time the probability of doom has reached the top of the media stack. It’s the eleventh time P(doom) has gone prime time since May 2023. In earlier cycles, AI-related stocks often moved sharply in the days surrounding these events, and many later advanced during broader periods of AI-driven investment and demand.

A case in point is Nvidia. Between P(doom) triggers spanning May 2023 and September 2025, Nvidia’s stock posted significant gains during the wider AI‑capex cycle. Nvidia returned 239% in 2023, 171% in 2024 and 38.9% in 2025, and topped a $5 trillion valuation in October 2025 — with a closing all-time high of $235.20 in May.

This month’s P(doom) event has produced a different market reaction. Shares of chipmakers and neoclouds have fallen, while hyperscalers — the companies driving capital expenditures in the AI trade — have risen. Analysts cited expectations that a slowdown in frontier model development could reduce near‑term compute demand and allow hyperscalers to digest existing capacity.

Why Doom Cycles Keep Dominating Headlines

The mass media’s intense focus on AI’s existential risk is nothing new. I’ve noted 11 such incidents since May 2023. These spikes in media attention tend to follow a similar trajectory:

  1. A respected insider defects from an AI purveyor or warns of a terrifying AI catastrophe
  2. Experts attach probabilities to AI’s likelihood of causing human extinction
  3. Cable news and newspaper front pages follow the story for four to seven days
  4. Legislators meet and fail to pass relevant laws
  5. The news cycle moves on

This month’s P(doom) event seems to be unfolding accordingly. First, “We are gambling with our lives” by racing toward superintelligence, noted a Sept. 8 resignation post by Jacob Coxon — which drew more than 110 million views on X.

Next, “We really do earnestly believe AI could kill all humans,” Anthropic’s alignment science lead Evan Hubinger, noted — adding his p(doom) topped 10% within a decade, according to CNBC .

Step three — the media swarm — has been underway for the last week, abetted by CEOs calling for an industry-wide slowdown. On Sept. 12, Anthropic CEO Dario Amodei published an essay calling for an industry-wide slowdown — which OpenAI CEO Sam Altman, Google DeepMind CEO Demis Hassabis and SpaceX boss Elon Musk largely endorsed, per Business Insider .

Sen. Bernie Sanders proposed a superintelligence ban , while President Trump called the conversation a "hoax," reported CBS News .

It remains to be seen how long the topic will remain on front pages and whether any legislation will follow. Historically, the media has moved on and no federal AI regulation has become law.

The table below summarizes eleven episodes, when they happened and what made each one the lead news.

How Markets Respond During Doom Cycles

Investors have taken a differentiated approach to AI stocks when P(doom) reaches the front pages. They evaluate which segments of the industry could face near-term pressure and which could benefit from shifts in demand or capital expenditures.

On Sept. 14, chipmakers declined while hyperscalers rose. Nvidia fell 2.78% to close at $212.22, AMD lost about 5%, Intel declined about 7% and the Philadelphia Semiconductor Index dropped roughly 6% — while Alphabet rose nearly 2%, Microsoft 1.6% and Meta 1.4%, noted Fortune .

Analysts said a coordinated slowdown in frontier model development could reduce demand for new compute capacity, affecting chip and neocloud providers. They also noted that hyperscalers could respond by reducing capital expenditures, digesting existing capacity and increasing free cash flow.

Between P(doom) triggers, Nvidia shares often advanced during broader periods of AI-related investment. Examples from Google Finance :

  • May 30, 2023 → Nov. 1, 2023: +15%
  • Nov. 1, 2023 → Nov. 17, 2023: +23%
  • Nov. 17, 2023 → May 17, 2024: +88%
  • May 17, 2024 → Sept. 27, 2024: +32%
  • Sept. 27, 2024 → Jan. 27, 2025: -1%
  • Jan. 27, 2025 → May 22, 2025: +9%
  • May 22, 2025 → Sept. 9, 2026: +71%

DeepSeek is a notable exception. Nvidia fell 17% and shed $589 billion of market value — the largest one-day loss for any company in history — with Broadcom down 17% and ASML down 7%.

Analysts attributed the drop to DeepSeek’s lower costs, which affected margin expectations. Broadcom finished 2025 up 49% and ASML rose 36% , per CNBC .

Analysts also noted that some AI-doom announcements have come from labs making significant compute commitments. For instance, Anthropic paired its extinction warning with a commitment to a gigawatt of Nvidia compute.

Implications For Investors

Analysts distinguish between competition‑driven shocks and doom-cycle headlines. DeepSeek’s lower costs affected margin narratives. In earlier cycles, human-extinction stories often coincided with short‑term volatility but did not consistently alter longer‑term performance across the AI sector.

Three developments could materially affect the AI trade, according to analysts: a 10% decline in chip stocks accompanied by legally enforceable restrictions on computing or permitting; an actual training moratorium; or federal legislation with enforcement power that restricts frontier AI.