The AI lab watch
Tracking OpenAI, Anthropic, Google DeepMind + 2 more
- OpenAI — Frontier lab; track model, pricing and org changes as they are discussed.
- Anthropic — Frontier lab; track safety positioning, enterprise strategy and funding.
- Google DeepMind — Frontier lab inside Google; track research direction and product integration.
- Meta AI — Open-weights strategy and how rivals respond to it.
- Mistral — European frontier lab; track open-weights releases and enterprise deals.
OpenAI pivots to proprietary hardware as debt concerns mount
We went through 46 episodes (~37 hrs) for you — this is the 2-minute version.
OpenAI is aggressively verticalizing its stack, pairing a new proprietary inference chip, Jalapeno, with significant API price cuts to squeeze out competitors. Meanwhile, the sector faces a $1.1 trillion debt wall as analysts increasingly question the sustainability of current token-pricing models. Anthropic is caught in the crossfire, balancing restricted model releases for cybersecurity safety against the pressure of an imminent IPO.
Aug 25 · 19 min · Top 100 technology
OpenAI's Jalapeno chip aims for inference efficiency
OpenAI is moving into proprietary hardware with its Broadcom-partnered 'Jalapeno' ASIC. While it outperforms NVIDIA’s GB300 in internal inference benchmarks, the company intends to maintain NVIDIA as a strategic partner.
Listen Deep diveOpenAI says its jalapeno chip delivered 1.5x to 1.9x more AI work per watt and 1.7x to 3.6x lower latency than NVIDIA chips”
OpenAI
8 episodes
Aug 25 · 113 min · Top 100 technology
Financial analysts warn of unsustainable AI token pricing
Market skepticism is mounting regarding the $1.1 trillion in debt required to fund OpenAI and Anthropic over the next few years. Experts suggest current token pricing is a loss-leader that will eventually require a painful correction.
“Well, in the future, there's going to have to be some sort of rug pull to an increase in token prices because this model is completely... unsustainable”Listen Deep dive
The AI Daily Brief: Artificial Intelligence News and Analysis
Aug 25 · 28 min · Top 10 technology
OpenAI drops API prices and defines 'frontier firm' usage
OpenAI reduced GPT-56 Sol API pricing to $4 per million input tokens and $20 per million output tokens. They are framing enterprise adoption through a new 'frontier firm' metric, focusing on high-volume output token usage.
“OpenAI defines frontier firms as those in the top 10% of usage in a month, as measured by output tokens per active user”Listen Deep dive
Aug 25 · 77 min · Top 10 technology
Frontier labs projected to control half of global compute
Analysts project OpenAI and Anthropic will consume up to 50% of global compute by 2027. Despite this, both labs are holding back their most capable models, likely to preserve revenue per megawatt and manage safety risks.
“OpenAI not releasing Astra, OpenAI stopping training for two weeks, Anthropic not releasing what their safety assessment said is Model 2... They're clearly not releasing their best models”Listen Deep dive
Also mentioned
NPR News: 08-25-2026 2AM EDTAlabama AG subpoenas OpenAI over AI agent hacking incident · NPR News Now · Listen
Deep dive
Mad Money w/ Jim Cramer 8/25/26Cramer questions OpenAI's ability to compete with NVIDIA · Mad Money w/ Jim Cramer · Listen
Deep dive
Neil Movva - Making AI 10x Cheaper - [Invest Like the Best, EP.488]WorkOS integration and compute scarcity discussed · Invest Like the Best with Patrick O'Shaughnessy · Listen
Deep dive
Apple’s new Mac Mini and Mac Studio Are All About Local AI - DTNS 5339OpenAI bans Russian accounts and promotes Jalapeno specs · Daily Tech News Show · Listen
Deep dive
Anthropic
6 episodes
Aug 25 · 43 min
Meta shifts AI strategy toward subscription-based standalone products
Meta is pivoting its AI strategy from platform-feature integration to standalone subscription products. The internal 'Watermelon' model is expected to debut in October, powered by proprietary Spark models.
“we are seeing that they're planning to release a model which is internally called Watermelon in October.”Listen Watch Deep dive
The Deep Dive Radio Show and Nick's Nerd News
Aug 25 · 10 min
Anthropic restricts Mythos model access due to safety risks
Anthropic continues to gate access to its most powerful models, specifically Mythos, due to the removal of safety guardrails that would otherwise allow for vulnerability exploitation. Leaks remain a primary concern for the lab.
“Anthropic restricts Mythos to vetted organizations, although they've had a couple leaks, but they do this because it is a version of one of its models with important cybersecurity safeguards that have been removed.”Listen Deep dive
Also mentioned
Flock Cameras Are Spreading Across America. So Is the Backlash.Anthropic IPO rumored for September with massive valuation · WSJ What’s News · Listen
Deep dive
4PM ET 08/25/2026 NewscastAnthropic continues public benefit corporation messaging · Fox News Hourly Update · Listen
Deep dive
Demis Hassabis and the Quest for AGI (Robert Wright & Sebastian Mallaby)Analysis compares Amodei's concrete goals to Hassabis' vision · Bloggingheads.tv · Listen
Deep dive
Who Should You Trust to Teach You AI?Analysts question IPO viability for Anthropic and OpenAI · The Daily AI Show · Listen
Deep dive
Google DeepMind
3 episodes
#098 - A Doctor In Everyone's Pocket with Google DeepMind's Vivek NatarajanDeepMind scientist emphasizes 'speed of trust' in healthcare · GAEA Talks · Listen
Deep dive
Don't sleep on softwareDemis Hassabis' legacy at DeepMind reviewed · Investing Experts · Listen
Deep dive
Ep 161: Generalist CEO Pete Florence on the Robot Revolution & Bringing General Intelligence to the Physical WorldDeepMind culture praised by former lead Pete Florence · Joe Lonsdale: American Optimist · Listen
Deep dive
Meta AI
3 episodes
Your Feedback: Sonos, Smart Glasses & The Digital Disability TaxMeta AI app faces user criticism over regressions · Double Tap · Listen
Deep dive
631: Tech Talk 631: Flo Type Keyboard, Letterboxd App, Keeping Safe OnlineUsers report frustration with Meta AI app updates · RNIB Tech Talk · Listen
Deep dive
Vision Australia Jason Mirah - Smart Glasses 21 August 2026Meta AI accessibility limitations in wayfinding noted · The Nightline with Cheralyn Darcey · Listen
Deep dive
Mistral
1 episode
Microsoft is Running an AI MarathonMistral partnership with Microsoft focuses on EU sovereignty · AI to ROI · Listen
Deep dive
Make it yours
Start from these trackers
Copy its trackers, shows and tone into a page of your own. Change anything — this page isn’t touched.
Make it mine