Frontier research. Frontier week.
from the editor
Is it just me, or was this week an unusually productive week for Bittensor?
Hello from San Francisco. It’s my first time in California (!), for Breakout and Bittensor 101. And, aside from being amazed at the popularity of DeAI and distributed systems here in the Bay Area, it feels great to be pitching Bittensor when TAO is on such a high.
Intel publishing a paper with Targon. Const appearing on not one but two crypto podcasts, @notthreadguy and @VirtualBacon. New product from Vidaio, new task suite from Score, and year-time-high for TAO above $350.
Live from the West Coast, I bid you: LFG 🚀 @macrozack
Chris Zacharia, Editor
Deep dive
AdTAO isn't a protocol-native team that learned PPC. It's a PPC team - with two exits, thirteen years of partnership, and a genuine data moat - that learned the protocol. The best problems don't always come from inside the ecosystem. Sometimes they come from operators like Rob, who spent years frustrated by a broken industry and just needed Bittensor to complete the picture. Bitstarter provided the bridge. "I've stepped into the universe," is how he put it. I believe him.
Highlights
This week in TAO
The kind of headline that, one year ago, would’ve seemed like pure fantasy: Last week: NVIDIA Inception. This week: Intel Incentivized. Bittensor’s core compute and AI subnets have achieved more in a couple weeks than most subnets envision in their entire roadmap. Targon has been building their Confidential TEE infrastructure for over a year; the paper demonstrates Targon and Intel’s provisioning of decentralised compute on untrusted hardware using Intel TDX and encrypted CVMs. But the optics are just as significant - and I mean that as a compliment. A Bittensor subnet has coauthored a whitepaper on Intel's website, announced by both parties, in the same week that Targon’s compute powered Quasar's long-context breakthrough.
When Const shares your paper, you're doing something right. SN24 Quasar has published new research taking aim at one of the hardest unsolved problems in AI: memory and attention at scale. The fact that it's landing ahead of anything comparable from the big labs is exactly the kind of thing that should make Bittensor bulls smile. This is what the protocol is supposed to produce - frontier research, incentivised from the bottom up, beating centralised incumbents to the punch. Huge congratulations to the Quasar team. More of this, please.
OK, stop for a second and actually think about what SN44 has achieved here. You download an app. It finds your cameras. You ask it questions - in plain English, like you're chatting to Claude. It runs inference locally on your hardware. No cloud pipeline, no egress costs, no latency nonsense. Just: how many people are in the room right now? What time does it get busiest? What are they picking up? The applications for retail alone are enormous. Security, workflow automation, accessibility - and that's before anyone gets creative. This is the kind of thing that, in five years, will seem so obvious that nobody will remember a time before it existed. It's on Bittensor. Built in the open. Accessible to anyone. Extraordinary. Big up @MaxScore, who’s been a huge credit to the protocol for a long, long while.
One of Bittensor’s most commercially viable subnets just proved its potential, announcing two revenue-generating partnerships, the first with NominoData and the second with an unnamed company who are tapping into Yanez’s biometrics product. What’s more, the revenue will return to SN54’s liquidity pool, which according to founder @josercaldera will reach 80T just from commitments signed so far.
Builder Spotlight
Every subnet starts as an idea. Most of those ideas used to die in a Discord thread at 11pm, half-formed, upvoted by three people and then forgotten forever. Bitstarter's new forum gives them somewhere to live: post your concept, gather upvotes and comments from the community, and the best ideas get pitched to a network of builders, backers, and advisors. I'm genuinely curious what floats to the top. If you've been sitting on a subnet idea - and I suspect some of you have a whole document full - now there's no excuse.
In the media
SN68 Metanova (drug discovery), SN93 Bitcast (decentralized creator marketing platform), and SN44 Score (specialised vision models) all appeared on the This Week in Startups podcast this week - 94,000 followers on X, a loyal audience of founders and early-stage investors. The host's framing was exactly right: three totally different problems, one common answer - decentralised compute reduces economic friction and lets you build faster.
A few months ago, one subnet mention on a show like this was news. Three on the same episode, in the same week that Targon is on Intel's website and Quasar is being shared across AI Twitter, feels like something clicking into place. I don't want to oversell it. But I'm not going to undersell it either.
Tweet of the Week
Jason casually triggers a mini bull run with the absolute lowest common denominator minimalist Bittensor maxi post ever.
Ready for it?
Here goes…
$tao > $btc
Chart of the Week
Macrocomos achieved its fastest-growing repo adoption ever, with its Dataverse Rust CLI for agent-first querying of real-time social media data. And they also did the same thing for SN1, Apex, launching an agent-first CLI.
look out for…
Breakout, Bittensor Commons pilot community event, hits San Francisco on Monday …plus Bittensor 101 at Frontier Tower. See you there ✌️
"It is not the strongest of the species that survives, nor the most intelligent, but the one most responsive to change."
- Charles Darwin
Bringing TAO to the world.