# The Man Who Calls BS On AI: They’re LYING About AI, 2027 Is When It All Breaks! | Ed Zitron

Source: https://www.youtube.com/watch?v=Lf5oqGOCRCM
Recap page: https://rapidrecap.app/video/Lf5oqGOCRCM
Generated: 2026-08-27T13:14:57.689+00:00

---
## The Gist

Generative AI is an unsustainable corporate con built on massive unprofitable data center spending, fake productivity metrics, and a looming market crash expected by 2027. Despite multi-trillion-dollar valuations and massive hype, AI companies are burning billions of dollars without clear paths to profitability.

## Quick Overview

Generative AI is an unsustainable corporate con built on massive unprofitable data center spending, fake productivity metrics, and a looming market crash expected by 2027. Tech critic Ed Zitron exposes how OpenAI, Anthropic, and major tech giants are artificially inflating the AI boom through massive capital expenditures while hiding their actual burn rates and lack of true revenue.

**Key Points:**
- OpenAI lost $5.09 billion in 2024 after bringing in $3.7 billion in revenue against $12.48 billion in total expenses, with $5 billion spent strictly on model training.
- Up to 70 percent of AI revenue across major players comes from companies funding each other, such as Amazon investing $50 billion into OpenAI and Google and Microsoft pumping billions into Anthropic.
- Data centers built for AI require staggering amounts of energy, with projects like Stargate Abilene in Texas demanding 1.2 gigawatts of power, which is 1.5 times more energy than the entire city of Bristol.
- Model Evaluation and Research data shows that grounded hallucination rates on simple summarization tasks for top models like ChatGPT and Gemini hover between 0.7 percent and 7 percent, proving they are expensive text-prediction machines prone to error.
- Venture capital firms and private equity have poured hundreds of billions of dollars into AI infrastructure despite venture capital historically experiencing a low return of 0.81 to 1.21 on invested dollars since 2018.
- Unlike the internet or mobile technology which eventually found organic consumer demand, generative AI is experiencing forced enterprise spending subsidized by tech monopolies gaming algorithms and attention economies.
- The looming market crash for the AI bubble is projected for 2026 or 2027 as capital expenditures outpace returns and major tech companies are forced to restate financial realities.

![Screenshot at 00:49: Ed Zitron details the staggering financial losses of OpenAI, revealing a $5.09 billion net loss driven by billions spent on model training.](https://ss.rapidrecap.app/screens/Lf5oqGOCRCM/00-00-49.jpg)

**Context:** Ed Zitron is a prominent tech critic, podcaster behind Better Offline, and founder of EZPR who has spent years warning that the artificial intelligence industry operates as a massive financial bubble. In this interview with Steven Bartlett on The Diary Of A CEO, Zitron dismantles the narratives pushed by Silicon Valley executives regarding profitability, job displacement, and economic growth.

## Detailed Analysis

Generative AI operates as a non-consensual technological push driven by ultra-rich executives and venture capitalists who are misleading the global economy. Companies like OpenAI and Anthropic rely on leaked annualised revenue run rates rather than audited financials, masking billions in actual cash burn and unsustainable operating costs. The underlying infrastructure requires trillions in capital expenditures for GPUs and data centers that consume more energy than entire cities, yet generate no organic economic growth. The hype surrounding AI replacing all human jobs or winning an artificial intelligence race against China is a marketing tactic designed to keep investors paying for expensive compute power. Ultimately, the artificial intelligence bubble will burst when funding dries up and companies can no longer subsidize their own unprofitable software.

### Topic 1: Generative AI as a Corporate Con

Ed Zitron asserts that generative AI is fundamentally a con sold by tech billionaires who overstate what models can do.

- Generative AI is sold as magical software that will replace all jobs and cure diseases, but it is actually expensive, unreliable, and unprofitable cloud software.
- Over 70 percent of AI revenues for companies like OpenAI and Anthropic come from other tech giants funding them in circular economic loops.
- Amazon invested $50 billion into OpenAI while Google and Microsoft poured billions into Anthropic to keep the ecosystem artificially afloat.

![Screenshot at 04:28: Ed Zitron presents the six leading AI companies and explains why their fundamental business models are a con.](https://ss.rapidrecap.app/screens/Lf5oqGOCRCM/00-04-28.jpg)

### Topic 2: The Staggering Energy and Cost of Data Centers

The hardware required to train and run large language models demands unprecedented energy and capital investments.

- AI data centers use specialized Nvidia GPUs that require massive bandwidth, memory, and electricity compared to traditional CPUs.
- The Stargate AI data center in Abilene, Texas spans 1,100 acres and requires 1.2 gigawatts of power, making it 1.5 times larger in energy consumption than the city of Bristol.
- Tech companies have spent over a trillion dollars on capital expenditures for data centers and GPUs with the expectation of making hundreds of billions in return, despite current unprofitability.

![Screenshot at 07:07: An aerial view of the Stargate AI data center in Abilene, Texas, showcasing the massive physical scale and energy requirements of modern AI infrastructure.](https://ss.rapidrecap.app/screens/Lf5oqGOCRCM/00-07-07.jpg)

### Topic 3: Hallucination Rates and Software Reality

Model Evaluation and Threat Research data proves that advanced language models still struggle with basic accuracy and reliability.

- Top frontier models like ChatGPT and Gemini exhibit grounded hallucination rates ranging from 0.7 percent to over 7 percent on simple summarization tasks.
- Ed Zitron emphasizes that these models are glorified text prediction engines rather than autonomous intelligences capable of reasoning.
- Using AI tools often requires complex prompt engineering and troubleshooting, making them less efficient than traditional software solutions for many professionals.

![Screenshot at 27:03: A chart displaying grounded hallucination rates for the top 25 large language models, highlighting error rates across the industry.](https://ss.rapidrecap.app/screens/Lf5oqGOCRCM/00-27-03.jpg)

### Topic 4: The Dot-Com Bubble Parallels

The current artificial intelligence boom mirrors the historical over-speculation and capital malinvestment of the 1990s dot-com era.

- During the dot-com bubble, investors poured billions into internet startups that had no profits before the market crashed in 2000 and wiped out trillions in wealth.
- Nobel prize-winning economist Paul Krugman famously predicted in 1998 that the internet's impact on the economy would be no greater than the fax machine.
- While the internet ultimately changed the world, the initial wave of overhyped companies collapsed because valuations detached completely from economic reality.

![Screenshot at 43:03: Archival footage of a horse-drawn carriage transitioning to early automobiles, illustrating how disruptive innovations eventually overtake older systems despite initial economic friction.](https://ss.rapidrecap.app/screens/Lf5oqGOCRCM/00-43-03.jpg)

### Topic 5: Job Disruption and the Myth of Autonomous AI

Corporate promises that AI will replace human labor are largely exaggerated marketing ploys designed to drive adoption.

- Media outlets and tech executives have spent years pushing the narrative that AI will take high-paying careers, forcing businesses to adopt tools out of fear of falling behind.
- Autonomous driving projects like Waymo experience edge-case failures in rain or complex traffic environments, proving that true autonomy remains far more difficult than promised.
- While AI assists with coding and text generation, it does not possess human contextual understanding or autonomous judgment.

![Screenshot at 63:08: A Waymo autonomous vehicle stuck in traffic in Las Vegas, highlighting the real-world operational limitations and edge-case failures of autonomous systems.](https://ss.rapidrecap.app/screens/Lf5oqGOCRCM/00-63-08.jpg)

### Topic 6: The Looming Market Crash in 2027

Without sustained external consumer demand and with venture capital returns dwindling, the AI industry faces an inevitable financial reckoning.

- Venture capital funds have experienced a total value return ratio between 0.81 and 1.21 since 2018, indicating that tech investments are failing to yield profitable exits.
- OpenAI delayed its initial public offering timeline as financial records reveal massive net losses driven entirely by model training costs.
- As tech giants like Microsoft, Google, and Amazon face contracting returns on their multi-billion-dollar AI investments, a broader market correction is projected to hit by 2027.

![Screenshot at 128:57: Ed Zitron points to the boxes representing AI labs and cloud providers, explaining why OpenAI running out of money will trigger a market-wide domino effect.](https://ss.rapidrecap.app/screens/Lf5oqGOCRCM/00-128-57.jpg)