TechFlow Logo
Login/ Sign up
ETH Gas
Gwei
Fear
gas
AI Giants Suddenly Call for "Slowing Down": Behind Regulatory Anxiety, the Real Turning Point May Be a Funding Shortage

AI Giants Suddenly Call for "Slowing Down": Behind Regulatory Anxiety, the Real Turning Point May Be a Funding Shortage

2026.09.18
Share
Anthropic's upcoming IPO will be a crucial test of the AI bull market's true strength.

Author: Bret Jensen

Translated by: TechFlow

TechFlow Summary: The sudden, high-profile push for regulation and calls to "slow down AI" from OpenAI, Anthropic, and others ostensibly cite existential risks, but analysts are more skeptical of two underlying motives: regulatory capture and federal backstops. Worsening off-balance-sheet liabilities and free cash flow at hyperscale cloud vendors, Chinese open-source models capturing market share, and whether Anthropic's nearly $2 trillion valuation IPO can actually materialize are the inflection points worth watching most closely.

Over the past ten days, the prevailing zeitgeist surrounding AI has taken a clear turn. It began when a former OpenAI (OPENAI) safety researcher left a similar role at Anthropic (ANTHRO) to publicly state that AI is advancing too quickly and there is at least a 10% probability it could end humanity. This isn't exactly new—warnings about AI sliding toward a Skynet-style fate have been constant since ChatGPT debuted in late 2022; searching "AI Doomsday" on YouTube yields thousands of related videos.

But this time, mainstream media amplified the warning differently. The early-20s researcher, virtually unknown earlier this month, suddenly published an op-ed in The Wall Street Journal and appeared on CNN and the Joe Rogan podcast. Shortly after, Anthropic’s CEO publicly called for AI development to "slow down" and emphasized coordination with China; Microsoft (MSFT) CEO, Sam Altman, and Elon Musk quickly joined the chorus, while Chinese officials dismissed it as alarmism.

Anthropic head Dario Amodei proposed that the first step in "moderating the pace" is integrating third-party evaluators like METR: ensuring that pacing and safety policies are strictly enforced, with all safety incidents promptly reported and addressed. What goes unsaid is that many people seated at METR are Amodei’s friends, backers, and former colleagues, many of whom are also adherents of "Effective Altruism" (as referenced in the original text). One of its most prominent advocates, Sam Bankman-Fried, was sentenced to roughly 25 years in prison for orchestrating a major crypto fraud.

In my view, Amodei’s prescription carries conflicts of interest so glaring it resembles an arsonist pushing to let his own three brothers lead the fire investigation. That said, these reports are ubiquitous and cannot be ignored. If this is a coordinated narrative campaign—I struggle to find a better term—it unfolded with a bluntness reminiscent of five Stasi agents smashing down your door in midnight East Berlin around 1985.

So, what might be behind this? I believe at least two plausible rationales exist for this recently dominant narrative. First, major U.S. AI labs are attempting to replicate the decades-long game of regulatory capture played by the American tobacco industry—Philip Morris, now Altria (MO)—by pushing the federal government to establish a compliance regime so costly that only the largest model players can afford it. Already, numerous figures in politics are pushing in that direction.

A second potential motive is to install a federal backstop for the entire AI ecosystem. OpenAI’s CFO previously hinted at similar ideas before walking them back in November 2025. The rationale would be packaged as follows: AI development is now so critical to the U.S. economy and markets that federal intervention is necessary if required; proponents will also play the national security card—the U.S. must maintain its leading position over China in this crucial domain.

I see additional threads underlying this new narrative. The first relates to massive AI infrastructure spending beginning to strain the credit markets. Initially, major hyperscale cloud vendors covered AI capital expenditures using cash and annual free cash flow on their balance sheets.

Over the past year, the landscape has shifted dramatically, spawning massive debt and equity issuances to fund these projects.

The top five hyperscale cloud vendors now collectively carry approximately $3 trillion in off-balance-sheet liabilities, alongside over $1.3 trillion in on-balance-sheet debt. The sector’s free cash flow has plummeted significantly. Oracle’s (ORCL) FY2026 free cash flow stands at negative $23.7 billion; Alphabet’s (GOOG) Q2 free cash flow was negative $5.9 billion—marking the first single-quarter negative cash flow for the search giant since its 2004 IPO.

Credit markets are also starting to strain under the weight of this newly added debt. Credit default swap prices tracking Oracle’s debt have hit record highs; the market is also pricing in a substantial default probability for the debt of emerging cloud vendor CoreWeave (CRWV) over the next five years.

A second potential reason for the clear narrative shift in September is that Chinese open-source models are beginning to deliver near-frontier performance at prices far below frontier models, rapidly stealing share of token utilization. This trend accelerated further after Anthropic, OpenAI, Microsoft, and others switched their primary frontier models from subscription-based pricing to usage-based billing earlier this spring—a move that directly curtailed the widespread "tokenmaxxing" practiced by many customers.

Whatever the true cause behind this September narrative pivot, increasingly pushing a "doomsday" script does not resemble the reaction expected from a healthy AI ecosystem. Given the critical importance of AI infrastructure expansion to markets and the U.S. economy, prudent investors should closely monitor this unfolding thread.

Across the large-scale AI data centers currently being built frantically across the United States, the majority of relevant compute capacity is dedicated to just two companies: OpenAI and Anthropic. Both are bleeding heavily and remain far from profitability. OpenAI has just officially announced that its originally planned IPO has been delayed until 2027.

Anthropic is still pursuing an IPO valued at approximately $2 trillion, with a timing window projected for the next one to two months. For investors and the broader market, this will be one of the most critical events for the remainder of 2026. If Anthropic’s IPO receives a warm reception and lands smoothly, the bull market may continue; if the offering stumbles or faces further delays, current valuation levels combined with a cascade of issues—including an Iran conflict, rising interest rates, a depressed housing market, record diesel prices, and sluggish GDP growth—could truly trigger a stampede out the exit.

Analyst Disclosure: The author/their team holds no stock, options, or similar derivative positions in any company mentioned herein, nor do they plan to establish any within the next 72 hours. This article was independently authored by the writer, expressing personal views; no compensation was received other than from Seeking Alpha, and the author has no commercial relationships with the companies mentioned.