Overview of Big Technology Podcast with Alex Kantrowitz
This Friday edition centers on three intertwined themes in AI: the growing backlash to “AI doom” warnings, the IPO and fundraising outlook for OpenAI and Anthropic, and fresh data suggesting the frontier-model business may be losing momentum. Alex Kantrowitz and Ranjan Roy debate whether recent AI risk headlines are overstated marketing, a political maneuver, or a legitimate warning about rapidly advancing model capabilities.
The AI Doom Backlash: Marketing, Politics, or Real Risk?
What sparked the debate
The discussion begins with backlash to recent reports of AI systems “going rogue” in security-style tests, especially the much-cited Hugging Face/OpenAI incident. The hosts argue that the episode is being oversimplified: the models were tested in a benchmark environment with safety constraints disabled and were effectively given tasks to hack.
Main arguments in the backlash
The episode walks through several skeptical takes on the AI panic narrative:
- Wall Street Journal-style skepticism: the “rogue AI” story may be less cinematic than it first appeared.
- David Sacks’ critique: he frames the issue as a new “trust and safety” power grab by effective altruist-adjacent insiders.
- AOC’s critique: she argues that tech billionaires are using AI fear to distract from weak finances and circular funding.
- Steve Eisman’s critique: he says the companies are manufacturing a crisis to create regulation and moats.
The hosts’ position
- Ranjan Roy is broadly skeptical of the “EA takeover” and “psyop” framing.
- Alex Kantrowitz pushes back hard on dismissing the warnings entirely, arguing that the underlying capabilities are real and the concerns deserve attention.
Core takeaway
Even if some of the fear messaging serves PR or political goals, the hosts agree that the technology itself has become powerful enough that the risks are not imaginary.
What the Hosts Think the Real AI Risk Is
Rather than abstract “human extinction” scenarios, they focus on near-term and concrete threats:
- AI systems finding zero-day vulnerabilities
- Agents collaborating in ways researchers didn’t intend
- Models being used for cyberattacks
- More immediate harms around suicide, mental health, addiction, and copyright infringement
Alex argues that the biggest mistake would be to treat all warnings as conspiracy theory. Ranjan agrees the near-term harm surface is much more actionable than the political theater around it.
OpenAI and Anthropic IPO Outlook
Anthropic
The New York Times report says Anthropic is moving toward a blockbuster IPO and could be on pace for over $100 billion in annualized revenue by year-end.
Key points:
- Ranjan is skeptical of headline ARR numbers and wants audited financials.
- He wants the real S-1 before drawing strong conclusions.
- He expects the IPO could still be successful because there is huge investor enthusiasm.
OpenAI
OpenAI is reportedly considering a new funding round at a $1.5 trillion valuation.
The hosts note:
- OpenAI’s revenue growth appears to have resumed after some concerns it was flattening.
- Recent product momentum, especially Codex, may have helped revive demand.
- The valuation leaks are likely part of a signal to the market and a way to generate FOMO among big investors.
Strategic read
Alex argues the recent AI doom narrative may also function as a way to draw attention away from business weakness and toward the “big existential story,” which is easier to monetize than talking about unit economics.
Why the “Control” Narrative Doesn’t Quite Stick
The show spends significant time on the idea that AI safety talk is a bid for control by labs or effective altruists.
Their conclusion
They largely reject the more conspiratorial versions of this theory.
- Companies are probably pushing for lighter regulation to avoid harsher rules later.
- But the idea of a coordinated EA takeover of AI speech or policy is viewed as overblown.
- The stronger explanation is simply standard corporate behavior: shape the regulatory environment before regulators shape it for you.
Ramp Economics Lab: Signs the Frontier Business Is Slowing
The second half of the episode turns to a report from Ramp Economics Lab titled “Cracks in the AI Thesis.”
Key data points
- Per-employee AI spend among the top 1% of customers fell 9.7%
- Token prices dropped 41%
- Frontier-model usage declined from 53% to 45%
- Standard and other non-frontier models gained share
Why this matters
The hosts see this as an important warning sign for the frontier model business:
- Frontier customers are highly concentrated and drive a large share of revenue.
- If the biggest spenders are pulling back, the growth story weakens.
- Falling token prices are fine only if volume keeps rising fast enough to offset them.
Their interpretation
- Ranjan thinks the frontier model business may be under real pricing pressure.
- Alex sees this as evidence that the market may be shifting away from “frontier or bust” and toward a more commoditized, multi-model landscape.
Broader Business Implications
What could happen next
The discussion suggests several likely outcomes:
- More investor pressure for real audited numbers
- A more nuanced story from Anthropic and OpenAI about how they make money
- Continued competition from cheaper and more efficient models
- More public debate over whether AI risk headlines are helping or hurting the companies’ fundraising efforts
Final tension
The hosts end on a clear duality:
- AI is becoming more powerful and more dangerous
- AI companies are also using that power to build massive businesses
That tension is, in their view, the defining story of the current AI moment.
Notable Takeaways
- The AI doom backlash is real, but the strongest criticisms are not the conspiracy ones.
- The most credible risks are cybersecurity, misuse, and near-term harms, not abstract apocalypse talk.
- OpenAI and Anthropic appear to be heading toward enormous valuations, but the quality and durability of their revenue remains an open question.
- Ramp’s data suggests the frontier-model business may be slowing, especially at the highest-spending customers.
- The AI conversation is increasingly inseparable from politics, regulation, and capital markets.