sun runner

starboy living the τao of poverty || thought journal

[sun runner @0xSunRun]

A good number of people in my mentions/DMs are looking for help taopilling the masses. In the interest of saving time, here is all you need to get started:

1.) If you only have time to do one thing, sit in on Bittensor co-founder @const_reborn's 1-hour lecture on Bittensor: https://t.co/86N9jFkb6d

2.) If your interest is sufficiently piqued, watch the best documentary of 2025 created by the @evert_scott and featuring all the usual suspects: https://t.co/ni4XtTH3M6

3.) If you now understand Bittensor at a high level and want to peel the onion a layer deeper without getting completely lost in the sea of complexity, check out @markjeffrey's of @stillcorecap's "State of TAO" report: https://t.co/sorBchlvEP

4.) Follow/sub to this Bittensor list I maintain that will be a portal into the ecosystem for you: https://t.co/22Q9eMy9zs

If these three things weren't enough to taopill you, feel free to slide into my DMs, and I'll get to you eventually (maybe).

Cheers.

[sun runner @0xSunRun]

>>>>"Another crypto category that is not dead is the "AI" category, full of high flying, fundamentally lacking, narrative driven tokens. The standout exception is Venice, a private AI platform with growing users and revenue, whose tokens are directly backed by the business rather than a narrative."

Kruger, like many large CT accounts, is completely misinformed (most likely due to a lack of effort) and, again, like many others, will be on the sidelines for the most important technological breakthrough in cryptocurrency since 2020.

[Alex Krüger @krugermacro]

I largely think of "crypto" as a failed asset class at this point.

I've written about the causes multiple times. Mainly, most crypto assets are worthless, or have dreadful value accrual, and most founders have abused the lack of guardrails and dumped on people indiscriminately, or are outright scammers.

On top of that we had the Memecoins SuperBullshitCycle, a trend that brought the worst out of people, and sucked everyone's souls & pockets dry. And then came the never-ending wave of DeFi hacks, which has dramatically increased since last April.

This can seem contradictory, as adoption of "crypto" is surging:

> Stablecoin adoption continues growing fast
> Politicians in the US are openly pro crypto
> Tradfi is looking at tokenizing everything
> Usage of equities & commodities perps is exploding in offshore and DeFi exchanges
> The US is in the early stages of adopting perps
> Prediction markets are becoming part of everyone's daily lives

These are more "blockchain" than "crypto", although there are some exceptions with a token in those fields, most of which have been performing very well in recent months. A few among those exceptions even distribute most revenue to holders via buybacks (Hyperliquid in particular), which is what every investor actually wants to see to be invested in a good business rather than a fleeting narrative.

We also have the privacy category. The one old school crypto category that is not liquid diarrhea. The world needs private non-custodial stores of value.

Crime in particular needs privacy, as proven by the DoJ confiscation of $15 billion in Bitcoin from Cambodia's pig butchering farms, legal filing for which was submitted on October 8, 2025 (coincidentally right before 10/10). Of course, everyone needs privacy, not just criminals, but crime flows are real, and large.

The asset attracting the most flows in this niche is Zcash. Zcash's recent performance has been fascinating, as it has been trending higher with bitcoin trending lower, a sign of real reallocation among bitcoiners.

Another crypto category that is not dead is the "AI" category, full of high flying, fundamentally lacking, narrative driven tokens. The standout exception is Venice, a private AI platform with growing users and revenue, whose tokens are directly backed by the business rather than a narrative.

So one could say old "crypto" is a failed asset class, but from the ashes come new beginnings, and the new face of crypto is one heavily dominated by the needs of Tradfi, prediction markets, AI, and privacy.

Crypto sucks. Long live crypto.

[const @const_reborn]

It’s a cyclical economy, rather than a hierarchical one.

And that will make all the difference

[sun runner @0xSunRun]

Few realize it yet but things have changed in a quite substantial way

[const @const_reborn]

Just orchestrated a 128 node permissionless decentralized training run, in 5 minutes, for 5 TAO, via @IOTA_SN9

They can do this up to 100B param models.
Unbelievable.

https://t.co/hJGZ6O5NrU

[sun runner @0xSunRun]

The corporate oligarchs are laughing as they talk about charging you thousands a month for the most important technology known to man

This is the opportunity DeAI is destined to exploit

[This Week in Startups @twistartups]

$20/month? Insane.

$200/month? No one will pay that.

$2,000/month? She doesn't know what she's talking about.

OpenAI CFO Sarah Friar on why everyone keeps getting AI pricing dead wrong:

@OpenAI, @thefriley, @Jason, @chamath, @theallinpod, @DavidSacks, @friedberg

[sun runner @0xSunRun]

@mewn21 @jords Open, permission-less, private, uncensored, cheaper

[sun runner @0xSunRun]

@const_reborn and @markjeffrey on stage from ~21:00 to ~52:00 - well worth the full listen. A few highlights that really got me excited (with some adlibbing from yours truly):

(1) In the agentic world we are hurdling towards, successfully mining a Bittensor subnet (given its difficulty) is the actual benchmark test for AI intelligence now.

(2) The Bittensor protocol, with its subnets, is the future of how organizations will coordinate/operate in an agentic world. A monetary-based set of rules of engagement to coordinate agentic intelligence and direct it towards a common goal is the most efficient system. All of the human elements of a company (think: HR, etc.) are superfluous in the agentic world. Const predicts that Anthropic will realize (if they haven't already) that this is the future and will head in this direction. Bittensor will have already been living in this world for 5-10 years at that point. If this prediction comes to pass, that clip in particular will stand out in history.

(3) Mark asked Const about his experiment with Arbos and his thoughts specifically on creating agents. Const responded that he has sidelined the Arbos project after having the realization that Open Claw, Hermes, Arbos, etc., are really just more software (in a world where software is now abundant and its moats are vanishing)

Instead, the real value is the underlying commodities that power the software. Those commodities are what Bittensor is designed to, and uniquely advantaged at, capturing, coordinating, and redeploying.

(4) Mark asked about other PoUW networks that have been popping up of late. Const thinks they are very bullish for Bittensor for two reasons. First, it broadens the overall DeAI category and brings more capital into the space. Second, using Pearl Network (PRL) as an example, it creates buyers for the digital commodities Bittensor is creating. In the case of Pearl, many of the miners rushing in to mine for PRL were renting their compute from Targon, Lium, etc. This is a symbiotic, not an adversarial, relationship.

[sun runner @0xSunRun]

@const_reborn and @markjeffrey on stage from ~21:00 to ~52:00 - well worth the full listen. A few highlights that really got me excited (with some adlibbing from yours truly):

(1) In the agentic world we are hurdling towards, successfully mining a Bittensor subnet (given its difficulty) is the actual benchmark test for AI intelligence now.

(2) The Bittensor protocol, with its subnets, is the future of how organizations will coordinate/operate in an agentic world. A monetary-based set of rules of engagement to coordinate agentic intelligence and direct it towards a common goal is the most efficient system. All of the human elements of a company (think: HR, etc.) are superfluous in the agentic world. Const predicts that Anthropic will realize (if they haven't already) that this is the future and will head in this direction. Bittensor will have already been living in this world for 5-10 years at that point. If this prediction comes to pass, that clip in particular will stand out in history.

(3) Mark asked Const about his experiment with Arbos and his thoughts specifically on creating agents. Const responded that he has sidelined the Arbos project after having the realization that Open Claw, Hermes, Arbos, etc., are really just more software (in a world where software is now abundant and its moats are vanishing)

Instead, the real value is the underlying commodities that power the software. Those commodities are what Bittensor is designed to, and uniquely advantaged at, capturing, coordinating, and redeploying.

(4) Mark asked about other PoUW networks that have been popping up of late. Const thinks they are very bullish for Bittensor for two reasons. First, it broadens the overall DeAI category and brings more capital into the space. Second, using Pearl Network (PRL) as an example, it creates buyers for the digital commodities Bittensor is creating. In the case of Pearl, many of the miners rushing in to mine for PRL were renting their compute from Targon, Lium, etc. This is a symbiotic, not an adversarial, relationship.

[Subnet Summer @SubnetSummerTAO]

Proof of Talk 2026 https://t.co/zBzUf54yaq

[sun runner @0xSunRun]

Hippius > Filecoin

[path.eth 🛡️ @Cryptopathic]

filecoin seems like a reasonable AI play, am I dumb

[const @const_reborn]

For this reason each subnet on Bittensor is intrinsically a force multiplier for others and collectively the stack of decentralized commodities natively integrate first with each other.

[sun runner @0xSunRun]

I mean chat, seriously, can it get anymore obvious?

[Garry Tan @garrytan]

Model routing is an important thing

Controversial idea: the frontier labs will want their AI harness to be the moat, but ultimately the best case for consumers is that model capabilities flatten and commodify

Preview of the AI Harness Wars of 2027

[sun runner @0xSunRun]

@notncssry Study harder

[sun runner @0xSunRun]

L1 gas tokens are an infinitely reproducible commodity with minimal locked in network effects and are going to 0 against tokens that have locked in network effects and adequate value capture

[Algod @AlgodTrading]

Crypto isn’t dumping because quantum risk

Its dumping because the tokens we send to tens of billions in valuation ended up delivering absolutely 0 value while in AI those funds are used much better

It’s our fault and have to do better, we need path to profitability instead of pretending governance tokens are any good

[sun runner @0xSunRun]

This

[Factory @FactoryAI]

Introducing model routing to Factory.

Factory Router picks the right model for every task, automatically.

Maintain frontier performance while cutting costs by 25%.

[sun runner @0xSunRun]

The year is 2028:

Your Mac boots up, and your agent comes online. On the agenda today: reconcile the quarter's books, draft three memos, debug last night's deploy, and plan the Tokyo trip.

The first three run locally on a 1T MoE trained on IOTA (SN9). No subscription, no API bill, nothing leaves the house.

The Tokyo planning needs more horsepower. Your agent could hit Claude's API, but it finds an 8T open model on Chutes (SN64) for 75% less and routes there instead. A few bucks in TAO.

A couple of hours later, and the work is done - time to head to the beach. While you're gone, your agent points the idle Mac at Actual (SN95), so it earns while you're gone.

Commodity intelligence: trained, served, and sold peer to peer.

[sun runner @0xSunRun]

The most important question in markets to be asking right now:

What happens when a distributed compute network, be it Bittensor, Pluralis, whoever, trains a multi-trillion parameter model?

[sun runner @0xSunRun]

@capradavis Imagine fading this team?

[sun runner @0xSunRun]

>he knows

[lbolord @lbolord]

you need to own the new means of production. own some mac minis or the new RTX spark in your home. plug them into actual computer (sn 95) and build an internet compute cluster with your friends that also own actual computers. then turn on earning mode on targon (sn 4) to passively monetize your cluster. connect your cluster to iota (sn 9) or pluralis to help train a 100B+ parameter model so you and your friends own a piece of a large decentralized model. then host the model on openrouter so you and your friends get paid to use the model

welcome to post agi

read the world for sale. own the new means of production

[sun runner @0xSunRun]

@DeFimorphosis

[sun runner @0xSunRun]

@BaznoCap Can I interest you in all of the above but add in that it’s a productive asset with accretive yield (buy + burn)

[sun runner @0xSunRun]

The year is 2026:

Targon announces the big boy version of the above.

Unreal. Massive.