The TAO Paradox: Why Bittensor’s Critics Might Be Right
From ‘Drug Deal Coin’ to Decentralized AI: How TAO’s Narrative Could Mirror Bitcoin’s Comeback
I remember first discovering Bitcoin in late 2017. This was right before the peak of the bull market, and I thought I was so late.
As an experiment, three of us childhood friends decided to invest together (note to self- this is always a bad idea).
My friend who originally introduced me to it, could not handle the volatility and after being up almost 50% on his investment in a matter of weeks he eventually sold at breakeven, never to buy any more again.
The other guy sold it all and spent his money buying ICO coins, everything from a supposedly gold-backed coin to a pot dealership and the funniest being a supermarket coin that promised to trace every food item back to its source and make you rich.
He even got phished for most of his Ethereum on a scam website before losing it all. He then wrote all cryptos off as a scam and hasn't touched it since.
Me? I held on for dear life. I devoured all everything I could find on Bitcoin, the books, papers, economic models, to the point that I knew more about it than the experts giving presentations on the topic in my company.
What this gave me was the belief to hold on to my investment, even in the face of extremely negative public opinion on the topic.
Conviction is key
What you really need to hold onto your investment is conviction to get you through hard times.
While Bitcoin is relatively mainstream now, even in 2017 it was an absurdity, an energy wasting Ponzi scheme and a scam to the wider public, good for nothing but drug deals and money laundering.
A lot of the spurious arguments on Bitcoin have now been debunked and I won't go into the details here, but at the time everything in the mainstream news was negative and if you took heed at all, you would not have touched it in a million years.
They say the richest Bitcoin investors are either dead or have lost their private keys, because they never had a chance to sell in the last 15 years.
Without having this, the best defense you have is to inform yourself as much as possible.
When you understand the asset and the counter arguments; you can hold in the face of negative news or public opinion and even add funds if you have the capital available.
I came across a post on Reddit this week outlining some counter arguments to Bittensor and found this as a good opportunity to challenge my assumptions to ensure I'm not too biased.
Most objections seem to trace back to two fears: whether incentives really stay aligned as the network scales, and whether open, permissionless AI can compete with future trillion-dollar companies like OpenAI.
Lets take these for a start and discuss.
Artificial Demand?
Bittensor Subnets are offering services at significantly subsidized prices to quickly onboard more users and increase overall traffic, despite significant losses at current market rates.
Is this sustainable?
I would say that subsidized pricing is an intentional go-to-market strategy.
In every networked tech market the cheap-or-free phase precedes monetization.
Think Google, Facebook, Twitter (X), the list could go on and on... Bittensor cofounder Jacob Steeves even notes that
“Large technology companies use the playbook of offering technology at a loss to control an industry and gain an advantage.”
The likes of Chutes (the biggest Bittensor subnet) are acquiring usage data and user relationships that later convert to paid endpoints. This is exactly what OpenAI did with ChatGPT.
All Bittensor subnets can do the same while they search for product market fit.
Can you really beat big tech?
OpenAI runs 200k+ GPUs; Bittensor aggregates much less than that, so would outputs be worse?
I tried estimating the number of GPU based on the number of miners in the major compute subnets (thanks to Taostats chatbot) and the numbers are miniscule in comparison (in the hundreds).
However, Bittensor’s design pushes miners to specialize: some subnets excel at research (SN1), others at fact-checking (SN70), image generation (SN19), drug discovery (SN68), AI agents (SN59) and on and on. The division of labor lets the network outperform monolithic models on niche tasks.
Also, incentives are continuous with daily rewards, not one-off Kaggle prizes, so models self-improve all the time.
As one response to the Reddit post said, instead of one giant supercomputer you have thousands of independent GPU nodes already working together through a marketplace.
The software stack and incentive mechanism are already proven with the first number of GPUs. To double the current number you just entice more miners and plug them in, no need to rewrite code or reorganize the database.
In other words, the scaling limit is your budget, not your engineering design.
This is a problem the big boys have that Bittensor does not need to worry about.
Is TAO inflation too high?
The argument here is emissions (release of tokens into circulation) start high to bootstrap the network, so skeptics fear perpetual dumping. However, this won't be the case for much longer.
TAO roughly follows the same 4-year halving curve as Bitcoin, causing emissions to drop geometrically and cap at 21 million. The first halving, currently projected for December 2025, will reduce daily issuance from 7,200 TAO to 3,600 TAO.
Staking is also frictionless, so most newly-minted TAO is immediately restaked instead of hitting exchanges, reinforcing network security.
Demand for TAO grows with every subnet because it is the only routing asset into Alpha tokens. Dynamic TAO ties subnet rewards to market demand for their Alpha tokens, creating a decentralized feedback loop.
Even though inflation remains a headwind until the first halving, the combination of emission decay, staking, and subnet-driven demand positions TAO to move to a deflationary or low emission state over time.
Couldn't someone pull a TRON and copy the code?
I always think of TRON when someone mentions forking code, adding a bunch of money and crushing the incumbent. This didn't happen with Bitcoin, Ethereum and its unlikely to happen to Bittensor now.
TRON initially operated as a token on Ethereum before launching its own blockchain in 2018. Launching in a blaze of publicity by its charismatic founder Justin Sun, it aimed to improve on Ethereum’s scalability and governance but has arguably achieved none of that.
TRON succeeded in niche markets (e.g., gambling and dApps) but has remained a marginal player compared to Ethereum’s scale.
Code is easy to replicate, community and Lindy effects are not.
Bittensor’s first-mover brand and grassroots culture are hard to replicate or supplant.
Subnets, liquidity and validator reputations are sticky network effects that a fork would need to recreate from zero while competing against an ecosystem already capitalized by Barry Silbert, OSS Capital, and multiple TAO treasury firms.
Where is the UX? This is all too complicated
This is a valid point, and something that rings closely with my own why for this newsletter.
Complexity hampers adoption and in no way can Bittensor be explained on the back of a napkin.
Bittensor is not meant for retail use as is, it is first and foremost a marketplace for machine learning work, not a finished consumer app.
Mining, validating, launching a subnet are developer workflows and earning TAO only becomes possible if you can write Python or deploy a GPU.
Bittensor needs better awareness and user-friendly front ends to unlock its full participant potential.
Right now, the main limiting factor of a decentralized network like Bittensor is just that people don't know about it.
And then once they know about it they don't know how to use it.
There's no "front door" that is an intuitive entry point.
This is something I am considering exploring in this newsletter in the near future, as my interests evolve in this ecosystem.
What does the "front door" look like for each of these Bittensor subnets, or even the Opentensor Foundation itself?
Back Yourself
In my mind Bittensor is at early maturity: protocol plumbing first, retail later.
Liquidity, UI wrappers, bridges, on-chain DeFi and consumer apps are being built, but they piggyback on the infrastructure the developers are creating right now.
Until that infrastructure properly exists, the risk/reward for a non-technical purchaser (like me) is mostly speculative price action, whereas developers can create immediate cashflow.
So as it is, I think of Bittensor as an asymmetric bet for the future. Sure there is a chance it might not succeed, but I think the odds favor and I am willing to back my conviction once again.
Hit reply with your own thoughts and tell me if I'm delusional. Either way I'll keep the coffee brewing and the code breaking.
Until next week.
Cheers,
Brian