Episode Description

The TAO Pod returns for Season 2 with a new format. Joseph Jacks, who co-hosted Season 1, is no longer part of the show — James addresses it directly in the opening minutes. There's no falling-out: JJ's attention has simply moved toward other areas of research, and as James puts it, people have different interests and move on to different things.

James notes he believes JJ remains invested in the ecosystem. Season 2 will run as a conversation series with guests from across the Bittensor world.

Hosted by James Altucher. In this episode, James is joined by Mark Jeffrey — early Bitcoin author (Bitcoin Explained Simply, The Case for Bitcoin), partner at Stillcore Capital, and host of the Hash Rate podcast — for the return of the TAO Pod after a several-month break. James and Mark discuss Jacob "Const" Steeves' rapid protocol changes and why both of them agree with all of them, Root Reborn ending the chain's daily dumping of subnet tokens and turning emissions into active subnet investment, the reports from Yuma and Sami Kassab arguing that a single breakout subnet takes TAO to $1,000 (and two or three takes it to $3,000–$5,000), why a general frontier model never eliminates the specialist layer (Score's computer vision and its PwC distribution deal), the Linux/Red Hat analogy for where subnet value actually sits, the collapse of the frontier moat as Kimi K3 matched Mythos and Fable within roughly a week, Engy serving Kimi K3 and GLM 5.2 at half the price of anyone on Earth by running on RTX 5090s near cheap power, Chutes and Parallax chasing a trillion-parameter mixture-of-experts model, Iota and Knyto attacking decentralized training from different angles, Actual Computer 95's private inference clusters and Nous Research's Hermes agent adopting them natively, Bitcoin as Earth's largest supercomputer that costs its owners nothing, why subnet tokens are rehypothecated TAO and drag TAO up with them, and the back-of-envelope math that puts TAO at roughly $2,000 off one subnet success out of 128.

Mark also notes this is not financial advice.

Key Timestamps & Topics00:00 Bittensor Value Thesis

00:46 TAO Pod Returns

01:29 Why Protocol Rules Change03:52 Bitcoin Versus Bittensor

06:09 Root Reborn Explained08:02 Breakout Subnet Flywheel

11:02 Can Claude Beat Subnets

12:37 Specialists Still Win

16:21 Inference Commoditization

18:04 NG Cheap Inference Edge

20:29 Prompt Data Flywheel

22:04 Chutes Parallax Training

23:36 Mixture Of Experts Path

25:03 Frontier Model Ambitions

25:56 Chutes Valuation Math

27:21 Decentralized Training Rivals

29:16 Actual 95 Free Inference

32:38 Bitcoin Economics Analogy

36:57 TAO Value Accrual

38:46 Subnet Versus Company Value

42:29 Token Flywheel Scenario

44:28 Crypto Access And Wrap Up