Many years back @KyleSamani and I had a call.
He dropped at 5 minutes because i said we used a 21M Bitcoin style supply curve.
Happy to pickup the conversation where we left off.
On the @theallinpod this week, @chamath asked @nvidia CEO Jensen Huang about decentralized AI training, calling our Covenant-72B run "a pretty crazy technical accomplishment."
One correction: it's 72 billion parameters, not four. Trained permissionlessly across 70+ contributors on commodity internet. The largest model ever pre-trained on fully decentralized infrastructure.
Jensen's answer is worth hearing too.
The first TAO halving has occurred.
- Current block reward: 0.5 TAO.
- Current daily emission: ~3,600 TAO.
- TAO halving reduces emission rate by 50%
- Halves: TAO emissions & subnet Alpha pool injections
- Unchanged: Alpha participant rewards (miners, validators + earned stake.)
TAO Synergies Raises Bittensor ($TAO) Holdings to over 54,000 Tokens, Largest Publicly Traded Holder of Top Decentralized AI Cryptocurrency. Read the press release here: https://t.co/0wXyJNzPSH $TAOX #DeAI#Bittensor#Crypto#AI
$TAO is the largest holding in the @Grayscale Decentralized AI Fund
Making up 33.53% of the entire portfolio
That's one third of the fund dedicated to #Bittensor
$TAO
.@opentensor $TAO helps leverage economic incentives and decentralized networks for open and accessible #AI development.
Grayscale Bittensor Trust is open for private placement for eligible accredited investors.
Learn more and see important disclosures: https://t.co/GN66eNEdSt
Institutions entering Bittensor $TAO and investing in subnets is a moment we’ve anticipated
Connecticut Innovations’ investment reinforces decentralized intelligence as a leading frontier
If Bittensor markets were fully developed AI startups, their combined market valuation could range between $100B - $200B+, competing with centralized Companies like OpenAI, Anthropic, and CoreWeave. That's what Dyanmic $TAO is about to do. Based on 64 Subnets, what happens when we scale to thousands? With 21 million design and halvenings!
Here is how:
I'll group these subnets into key AI industry sectors that align with their functionality.
▫️AI Model Development (Foundation Model Providers)
•Comparable Companies: OpenAI, Anthropic, Mistral, Hugging Face
•Subnets: SN 9, SN 29, SN 39, SN 11, SN 21, SN 37, SN 38, SN 3, SN 5, SN 35, SN 56, SN 58
•Market Valuation Estimate: $50B+ (OpenAI alone is estimated at $90B)
•Subnet Value Range: $100M - $3B each, depending on specialization
▫️Generative AI (Text, Image, and Audio)
•Comparable Companies: Midjourney, Stability AI, OpenAI (DALL-E, Whisper)
•Subnets: SN 4 (Targon), SN 19, SN 2, SN 54, SN 17, SN 18, SN 23, SN 26, SN 50, SN 46
•Market Valuation Estimate: $30B+
•Subnet Value Range: $100M - $2B
▫️Agentic AI (Autonomous Agents & AI Assistants)
•Comparable Companies: Adept AI, OpenAI (GPT Agents), AutoGPT, Rabbit R1
•Subnets: SN 1 (Apex), SN 45, SN 62 (AgentTao), SN 59, SN 20, SN 10
•Market Valuation Estimate: $20B+
•Subnet Value Range: $500M - $2B
▫️Predictive AI & AI Trading
•Comparable Companies: Numerai, PredictIt, Jane Street (AI-powered HFT)
•Subnets: SN 6 (Infinite Games), SN 30 (BetTensor), SN 41 (Sportstensor), SN 28 (Foundry S&P 500 Oracle), SN 44 (Score Predict), SN 48 (NextPlace AI), SN 57 (Gaia)
•Market Valuation Estimate: $10B+
•Subnet Value Range: $200M - $1.5B
▫️AI Compute Providers
•Comparable Companies: CoreWeave, Lambda Labs, Together AI
•Subnets: SN 51 (Celium), SN 64 (Chutes), SN 12 (Horde), SN 27 (Neural Internet)
•Market Valuation Estimate: $15B+
•Subnet Value Range: $500M - $5B each
▫️AI Data Pipelines & Data Labeling
•Comparable Companies: Scale AI ($14B), Snorkel AI ($1B), Labelbox ($1.3B)
•Subnets: SN 24, SN 13, SN 33, SN 40, SN 42, SN 52
•Market Valuation Estimate: $10B+
•Subnet Value Range: $300M - $2B
▫️AI Components & Tooling (Model Optimization, Deployment, Fine-tuning)
•Comparable Companies: Hugging Face, MosaicML ($1.3B)
•Subnets: SN 47, SN 31, SN 15, SN 49
•Market Valuation Estimate: $8B+
•Subnet Value Range: $300M - $1.5B
▫️AI-Powered Tools (Specialized AI Applications)
•Comparable Companies: Quora (Poe), Perplexity AI, BloombergGPT, DeepL
•Subnets: SN 34, SN 61, SN 32, SN 53, SN 8, SN 14, SN 22, SN 54
•Market Valuation Estimate: $8B+
•Subnet Value Range: $200M - $1.5B
▫️Scientific Research (DeSci)
•Comparable Companies: Protocol Labs, Biotech-focused AI startups
•Subnets: SN 25, SN 43, SN 36
•Market Valuation Estimate: $5B+
•Subnet Value Range: $200M - $1B
▫️General Infrastructure (Decentralized AI Infrastructure)
•Comparable Companies: Bittensor itself, Akash Network
•Subnets: SN 16 (BitAds), SN 7 (SubVortex)
•Market Valuation Estimate: $5B+
•Subnet Value Range: $300M - $1B
Step 2: Summing the Estimated Market Value
Adding up the projected values from each category:
While this is a broad estimate, AI markets often have overlapping applications, so a reasonable adjusted valuation would be:
Low estimate: $100B (accounting for overlap and market competition)
High estimate: $200B+ (if the network captures significant adoption)
▫️Valuation Per Subnet
•Top-tier subnets (major LLMs, compute, predictive AI): $2B - $5B+
•Mid-tier subnets (agentic AI, data pipelines, AI tools): $500M - $2B
•Smaller specialized subnets (niche AI markets, early-stage growth): $100M - $500M
This analysis shows the incredible potential upside of $TAO, as these markets are still in early development stages. If adoption accelerates and these AI markets keep growing at their current pace, Bittensor could capture a massive share of decentralized AI's future market.
Introducing Yuma Asset Management.
We’re unlocking access to decentralized AI through Bittensor subnet tokens.
Backed by a $10 million anchor investment from DCG, Yuma Asset Management bridges investors to a new asset class.
Learn more: https://t.co/PB9TKtWlqo