People Are Underestimating TAO (By DreadBongo)

DreadBongo posted a thought-provoking piece about Bittensor and how people are still underestimating it. We’ve made an article version of the tweet.

Explore below:

Let’s talk about something massive that’s flying under the radar: Bittensor. I genuinely believe people are sleeping on the sheer scale of what this decentralized intelligence network is already achieving—and where it’s headed. The numbers don’t lie, and the subnets within Bittensor are pulling off feats that rival, and sometimes straight-up beat, the biggest centralized AI giants in the game. And they’re doing it cheaper, faster, and with a scalability that’s frankly mind-blowing.

Let’s break it down.

Subnets Are Outpacing the Big Dogs

Bittensor’s subnets are already delivering results that compete with trillion-dollar behemoths—and they’re doing it with pocket change by comparison.

Take Ridges AI (SN62), for example. In just weeks, they hit 80% accuracy on SWE-Bench, a benchmark for coding tasks, outperforming Anthropic (you know, the folks behind Claude, valued at a whopping $183 billion). Ridges did it with less than $1 million spent. Meanwhile, Anthropic’s burning cash like it’s kindling. That’s not just a win—it’s a statement.

Then there’s Chutes (SN64), processing trillions of tokens every month while undercutting giants like Google Cloud by 90%. Yeah, you read that right—Google. The same Google that’s synonymous with “biggest tech.” Chutes is doing what they do, but leaner and meaner.

Or look at Macrocosmos AI. They’re pretraining foundation models on datasets that rival OpenAI’s, but at 1/10th the energy and cost. That’s not just efficiency; that’s a paradigm shift.

And don’t sleep on Templar (SN3), enabling continuous LLM training with 50x better gradient sharing, outpacing Meta’s Llama without needing Zuckerberg’s $100 billion+ infrastructure budget.

Even in biotech, Nova (SN68) is accelerating drug discovery through virtual screening on global compute, going toe-to-toe with DeepMind’s AlphaFold—but at a fraction of the cost. These subnets aren’t just competing; they’re redefining what’s possible.

The $TAO Flywheel

Here’s where it gets juicy: every subnet’s success pours rocket fuel back into the $TAO ecosystem. As subnets grow, they drive demand through staking, delegation, and emissions.

The math is simple—more wins, more value, more $TAO flowing through the network. Right now, top subnets like Chutes are valued around $80 million, or $650 million FDV. Ridges is climbing to similar heights. But let’s be real: when you stack their adoption, results, and raw efficiency against the billion- and trillion-dollar giants they’re outpacing, these valuations are laughably underpriced.

Think about it. Single subnets are doing what companies worth hundreds of billions do, but better and cheaper. It’s not a stretch to imagine wildly successful subnets hitting valuations in the tens of billions—maybe even $100 billion+ in the future. Now zoom out: Bittensor’s got 128+ subnets, all scaling, all compounding value, all feeding into the same ecosystem. That’s not just a network; that’s a decentralized intelligence juggernaut that could be worth trillions.

The Bigger Picture

With only 21 million $TAO ever in existence and emission halvings every four years, the supply is tight. Meanwhile, subnet growth is accelerating every single day.

The upside here is massive—way bigger than most people are pricing in. Sure, it’s early. Sure, it’s messy. And yeah, it’s volatile as hell. But you know what? That’s exactly what Bitcoin looked like in 2013. Back then, people called it a gamble, a fad. Now we can’t imagine a world without it.

Bittensor’s subnets are proving they can hang out with the big boys—Anthropic, Google, OpenAI, Meta, DeepMind—and beat them at their own game. As these subnets keep scaling, the value they create will flow back into $TAO, creating a feedback loop that’s hard to overstate. This isn’t just a project; it’s the foundation for a decentralized intelligence network that could redefine how we build and interact with AI.

So yeah, it’s early, and yeah, it’s wild. But if you’re not paying attention to $TAO and Bittensor’s subnets, you’re missing the forest for the trees. This thing’s got the potential to be unstoppable.

$TAO