Tech Brew Ride Home – "Tipping Point In The AI Horse Race?"
Host: Brian McCullough
Date: April 7, 2026
Duration: ~15 minutes
Episode Overview
This episode dives into a potential tipping point in the AI industry. Host Brian McCullough explores Anthropic’s explosive growth in revenue and its landmark deals with Google and Broadcom, considers the emerging "world model" race in AI, reflects on how AI has disrupted SEO, and issues a stark warning about quantum threats to cryptography. If you want an up-to-the-minute snapshot of Silicon Valley's biggest technology shifts, this episode delivers.
Anthropic Surpasses OpenAI: A New Leader in AI?
[00:34–06:40]
-
Anthropic's Blockbuster Growth
- Anthropic sealed agreements with Google and Broadcom for massive access to TPU (Tensor Processing Unit) capacity.
- According to Bloomberg reports:
- Anthropic’s revenue run rate now exceeds $30 billion, up from $9 billion just a few months prior.
- More than 1,000 business customers now spend over $1 million annually on its Claude services – a number that has doubled since February.
Notable Quote:
"They reported $19 billion in revenue run rate at the end of February. The math on that would suggest they added $10 billion in ARR in a month."
— Brian McCullough [02:37]- Anthropic now appears to have overtaken OpenAI in both revenue growth and potentially in total revenue—moving beyond being “the fastest grower” to being the current market leader.
Notable Quote:
"Anthropic could be profitable at this rate by the end of next year."
— Brian McCullough [03:05]Social Media Hot Takes:
- Yucheng Jin (on X) credits Anthropic’s success to focus on coding models and enterprise contracts ("Once you're locked into a year long contract, switching ... isn't easy. Claude Code shipping velocity is insane too.").
- Connor Sen wonders: "What is the rationale for valuing OpenAI above Anthropic? Anthropic should be $1 trillion at this point." [04:07]
-
Implications of TPU Deal
- Anthropic’s new hardware partnerships reduce its reliance on Nvidia by working with Broadcom to develop chips based on Google’s TPUs.
- This strategic collaboration gives Anthropic about 3.5 gigawatts of compute power starting 2027, pending continued commercial success.
- Broadcom’s CEO projects AI chip sales could top $100 billion next year, marking Broadcom as a bigger Nvidia competitor.
Memorable Moment:
"So will we look back on April 6 as the day we knew the tide had turned and Anthropic became the clear leader in the AI horse race?"
— Brian McCullough [05:16]
The “World Model” AI Land Grab
[06:40–10:46]
-
Next-Gen AI: Beyond Text and Language
- The host frames a new frontier: "world models," meaning AI systems trained to function beyond language and code, aiming for real-world understanding.
Notable Quote:
"But the dream of the thing that could get you all the way to real AI is still out there, and it’s world models."
— Brian McCullough [06:47] -
Project Prometheus: Bezos & Top Talent’s AI Ambition
- Jeff Bezos’ secretive company (“Project Prometheus”), led with ex-Googler Vikram Bajaj, has poached Kyle Kosich, ex-cofounder of Elon Musk’s xAI.
- Project Prometheus is aggressively recruiting engineers and researchers, especially those experienced in building massive infrastructure.
- The aim: AI that operates in the physical world (industrial automation, engineering, aviation, etc.), trained on unprecedented domain-specific datasets, e.g., jet engines.
- Prometheus plans to build a Berkshire Hathaway-style holding company, investing in industrial firms and embedding AI-trained forward-deployed engineers.
Notable Quote:
"The company is particularly focused on the industrial sector. It envisions a model that can understand the laws of physics and is trained on data from specific domains such as jet engine design."
— Brian McCullough [08:38]- Real-time data ambitions: Some AI startups are planning to launch satellites to train AI models with real-world, real-time physical data.
Memorable Moment:
"We don’t live in sentences and concepts. We live in a real Newtonian universe. So what if the next big thing is AI trained on all of that? ... The universe. The real world."
— Brian McCullough [09:45]
The SEO Gold Rush and AI Search Disruption
[13:02–15:00]
-
How AI is Upending Traditional SEO
- Google’s AI-generated overviews are accurate “nine out of ten times,” but with 5 trillion searches annually, that translates into tens of millions of errors every hour.
- A significant portion of these supposedly accurate AI answers are “ungrounded”—they cite sources that don’t truly support the provided information.
Notable Quote:
"With Google processing more than 5 trillion searches a year, this means that it provides tens of millions of erroneous answers every hour."
— Brian McCullough [13:17] -
SEO Industry Reeling
- SEO professionals are now desperately experimenting with tactics to get brands mentioned by AI search/chatbots, often through questionable means (“recommendation poisoning”).
- Ambiguity and hype reign: “Answer Engine Optimization. Generative Engine Optimization. Generative Search Optimization … endless new monikers.”
- The industry has seen massive traffic declines for tech publications—some have lost over 90% of clicks from Google, attributed to AI answers, changes to Google search, and user migration to chatbots.
Notable Quote:
"These AI-pilled SEOs that are saying we can do geo, we can do AIO. They are setting a dangerous precedent that they can influence AI in ways that are simply not true and ... you’re just setting yourself up for failure."
— Brittany Muller (quoted) [14:36]- Major anxiety about how success and reach will be measured in the new AI search world:
"We are just grasping at straws."
— Brittany Muller [14:14]
Urgent Quantum Computing Warning: Cryptography May Fall Sooner
[~15:00–End]
-
Quantum Threat Closer Than Expected
- Google researchers warn that quantum computers able to break elliptic curve cryptography (ECC—including crypto wallets and more) might require 20x fewer resources than previously believed.
- There’s now a credible fear that a “cryptography-relevant” quantum computer (CRQC) could exist by 2029—years earlier than most had planned for.
Notable Quote:
"The bet is not ‘are you 100% sure a CRQC will exist in 2030?’ The bet is ‘are you 100% sure a CRQC will NOT exist in 2030?’"
— Brian McCullough (paraphrasing Filippo Valsorda) [16:19]- Experts stress the "risk of inaction is now unacceptable." Migration to post-quantum cryptography must begin immediately to keep user data secure.
Final Thoughts:
"The job is not to be skeptical of things we’re not experts in. The job is to mitigate credible threats and there are credible experts telling us about an imminent threat.… We don’t have the time if we need to be finished by 2029 instead of 2035."
— Brian McCullough [16:54]
Key Takeaways
- Anthropic is exhibiting unprecedented revenue growth, possibly overtaking OpenAI and signaling a new order in the AI "horse race."
- The competition to build “world models” — AIs that can understand and operate in the physical world — is sparking a second AI gold rush and major talent migration.
- The explosion of AI-generated search results has disrupted the SEO industry, causing confusion, dubious tactics, and major declines in site traffic.
- Advanced quantum computing could upend cryptography much sooner than previously feared; urgent migration to post-quantum protocols is critical.
Notable Quotes Recap
- "Anthropic could be profitable at this rate by the end of next year." – Brian McCullough [03:05]
- "World models are something to be on the lookout for … the real AI dream is still out there." – Brian McCullough [06:47]
- "We don’t live in sentences and concepts. We live in a real Newtonian universe. ... What if the next big thing is AI trained on all of that?" – Brian McCullough [09:45]
- "These AI-pilled SEOs ... are setting a dangerous precedent." – Brittany Muller [14:36]
- "The risk of inaction is now unacceptable." – Brian McCullough [16:39]
For listeners seeking a fast-paced but nuanced summary of this week’s biggest trends in AI and tech, this is an indispensable episode.