TBPN: Cloudflare Outage, iMessages in Gemini 3, 𝕏 Reactions to Gemini 3 | Diet TBPN
Date: November 19, 2025
Hosts: John Coogan, Jordi Hays (plus Tyler & Ethan as panelists)
Theme: The intersection of cloud infrastructure, messaging, AI advances (especially Gemini 3), and reactions to major tech events
Episode Overview
This episode of TBPN explores the centrality of iMessage in personal and professional lives, the potential for AI assistants like Google’s Gemini 3 to integrate or interface with closed messaging ecosystems, and the far-reaching impact of recent outages and infrastructure decisions within the tech industry. The conversation moves fluidly across the latest AI benchmarks, industry intrigue (like Google’s internal drama), hardware monopolies, energy and environmental discussions, and everyday biz/podcast life, all in the hosts’ characteristic irreverent and conversational tone.
Key Discussion Points & Insights
1. iMessage as the Personal ERP: Integrating with AI Assistants
- iMessage's Role: Tyler considers iMessage as his "personal ERP system" and "data lake," central for both personal and business communications.
"It's my single pane of glass... the system of record for my personal life." — Tyler [00:55]
- The Power User's Perspective: Despite iMessage's centrality, Tyler and John note Apple's lack of motivation to build for 'power users,' making it unlikely they'll enable Gemini-like LLMs to utilize iMessage data.
- Apple-Google Dynamics:
"Apple is... paying Google to white label... Gemini in the next version of Apple intelligence and they're just going to be focused on integrating it within their ecosystem deeply." — John [02:56] Cross-platform portability is deemed unlikely because of these walled garden approaches.
- Use Cases Imagined: The idea of Gemini using iMessage contextually for drafting messages or suggesting gift ideas is discussed, but privacy and ecosystem segmentation will likely prevent this.
2. The Future of Messaging in the "AI Agent" Era
- Multiple Systems, No Unifier: Ethan highlights the reality that everyone juggles multiple messaging platforms (iMessage, WhatsApp, Signal, DMs), and true unification remains elusive.
- AI Agents as Intermediaries: The hosts imagine a near future where AI agents handle logistics and communication autonomously, possibly bypassing traditional messaging flows:
"I have an agent. My agent talks to Jordy's agent. They sort everything out... done completely separately from iMessage." — Ethan [04:47]
- Speculative Robot Assistant: Tyler jokes about giving a humanoid robot his phone to respond to every message, blurring boundaries between data silos.
3. Model Benchmarks and Google’s Strategic Moves
- Gemini 3 Pro vs GPT-5: The group discusses Gemini 3 Pro surpassing human professionals at GeoGuessr, raising benchmarking questions and concerns about overfitting, given Google's access to massive training data.
- Model Tiers Explained:
"3 Pro is like 5 thinking and then 3 flash... If that comes out, that will be instant." — Ethan [06:59]
- Model Tiers Explained:
- Google Antigravity IDE Drama: Google’s new developer platform allegedly forked much of its code from Winsurf (now part of Cognition) without even removing all old branding, illustrating classic big-tech acquisition quirks.
4. Hardware Infrastructure & Market Power
- Google’s TPUs vs Nvidia:
"Best model ever created from a benchmark standpoint. Didn’t use Nvidia chips, which are supposed to be a monopoly." — John [08:06]
- Market Implications: Even though Google makes powerful TPUs, if they don’t sell them to outsiders, Nvidia retains practical monopoly power—a “weird dynamic” that frustrates the hosts.
- Quote: "There isn't enough compute in the world as we all know for everything that we want to do." — quoting DeepMind’s Demis Hassabis per John and Tyler [09:45]
5. Naming and Merging AI Labs: Gemini’s Origins
- Google's internal drama and the merging of Brain and DeepMind is acknowledged as the namesake for "Gemini":
"Gemini is a reference to two formerly distinct labs, Google Brain and DeepMind, that were merged into one lab." — John [08:57–09:07]
6. Data Center Arms Race, Global Partnerships, and AI’s Resource Appetite
- XAI Data Center in Saudi: Musk’s XAI and Nvidia plan a massive 500MW data center in collaboration with Saudi partners—seen more as an infrastructure play than purely AI model training.
- "Elon is clearly... very good at large scale physical infrastructure build outs." — John [13:29]
- Resource Scarcity Issues: AI’s consumption of energy and water draws environmental scrutiny; the hosts dismantle a factual error regarding water usage in a popular AI book, underscoring the importance of accurate data.
7. Real Estate Antics: The Haunted Soil Machine
- Memorable Story: The hosts recount an absurd property viewing where a mysterious, humming box dubbed the “machine that cleans the soil” is discovered. The real estate agent’s evasiveness and the ultimate reveal ("machine shop from 100 years ago” contamination) provide comic relief.
“The more you tell me not to worry about it, I kind of wanna know more.” — John [18:28] “There was a laundromat here ago...” — Tyler [19:15] “No, it was a machine shop.” — John [19:51]
8. Cloudflare Outage: The Fragility of the Modern Internet
- Incident Recap: Cloudflare’s outage disrupted many major sites (X, ChatGPT, DoorDash, MTA, IKEA, etc.), exposing the extent of their infrastructure’s reach.
- “When your product goes down and then many of the services... also go down, it’s even more stressful.” — John [24:34]
- Reputational Irony: Outages unintentionally reinforce Cloudflare’s importance, as reliance becomes apparent at every layer of tech.
9. ADHD Telemedicine Scandal
- Discussion of a founder found guilty for Adderall overprescription via a telemedical startup:
“Telemedicine psychiatry startups have driven an unprecedented wave of amphetamine abuse.” — Tyler [25:11]
- Investors & Incentives:
"Doctors were a bug, not a feature." — John [26:23]
- Concern over replacing thorough medical review with conversion-optimized pill mills.
10. Economic Signals & Protein-Maxi Jokes
- AI Investment Boom: New $100B Brookfield/Nvidia/Kuwait AI infra partnership announced; a continuing tide of massive funding news.
- Light-hearted Moment: The “bowl economy” is benchmarked not by revenue, but by grams of protein per serving—a running joke on the show.
“This is the most important benchmark in the bowl economy, which I’m a huge fan of. But are we seeing acceleration? I want to be seeing 200 grams of protein, then 1,000, then 10,000...” — Tyler [22:53]
Notable Quotes & Memorable Moments
- On Apple and Google AI:
- "Some of the best walled gardens of all time." — John [02:08]
- On the future of AI agents:
- "If you have enough humanoid robots, though, then land is actually not that hard to..." — Ethan [15:08]
- On market dynamics:
- "It must be very frustrating... you do have two very clearly performant products that are not actually driving down cost." — Tyler [08:22]
- Comic Relief:
- "She said, that's just the machine. That's just the machine." — Tyler [18:14]
- On Cloudflare’s omnipresence:
- “It’s a great way to tell the world that the entire world runs on Cloudflare." — John [24:34]
- On rampant diagnostics:
- "Doctors were a bug, not a feature. Why waste time talking to a doctor just to get the medication you want..." — John [26:23]
Timestamps for Segments
| Timestamp | Segment/Topic | |:----------:|:------------------------------------------------------------| | 00:00–02:56 | iMessage’s role, why Apple/Google won't enable portability | | 02:56–04:47 | Practicality and privacy in connecting LLMs to messages | | 04:47–05:40 | The dream of AI agent communication | | 05:40–07:09 | AI models: GeoGuessr, benchmarking, model behavior | | 07:09–08:22 | Google Antigravity IDE and hardware choice (TPUs/Nvidia) | | 08:22–10:36 | Market dominance, org drama, and compute constraints | | 12:27–13:51 | Nvidia’s earnings, XAI’s data center ambitions | | 15:16–20:21 | The ‘haunted’ office tour (soil-cleaning machine story) | | 21:09–22:41 | AI’s energy/water footprint, debunking dramatic claims | | 23:26–24:34 | Massive AI investments & Cloudflare outage | | 25:11–27:15 | ADHD telemedicine and accountability | | 27:22–End | Quick notes on earnings, predictions, end-of-show banter |
Tone & Takeaways
This episode maintains an energetic, sometimes sarcastic tone, blending serious discussion of AI, cloud infrastructure, and corporate strategies with personal stories, in-jokes, and asides. The hosts emphasize how tech giants’ decisions reverberate throughout the ecosystem, often with unintended consequences or ironies (like outages revealing infrastructure linchpins), and poke fun at industry obsessions (from protein-fueled lunches to never-ending AI investment news).
For listeners:
If you want a behind-the-scenes look at how the big AI/tech sausage gets made—complete with quips, the real-world impact of corporate drama, and the kind of banter you’d overhear in Silicon Valley’s break rooms—this episode delivers.
