Key Highlights:
- Bittensor subnet operators have secured commercial agreements in computer vision, cybersecurity, GPU computing and AI inference.
- Subnet revenue increased from almost nothing over 18 months to a recent estimate of more than $32 million, according to Malanga.
- Bittensor uses validator scoring and Yuma Consensus to distribute $TAO rewards, while commercial demand remains the key test of lasting value.
Bittensor adoption expands through commercial AI services
Malanga presented commercial agreements involving Bittensor subnet operators as evidence of demand for services created through the network’s AI competitions. He also highlighted the growing number of operators purchasing their subnet tokens on the open market, arguing that Bittensor’s structure links demand for those assets with demand for its native token, $TAO.
The examples span computer vision and cybersecurity. Score is preparing to monitor Avia carwash locations, while RedTeam is working with financial institutions on transaction protection. Malanga said customers using RedTeam’s services collectively cover more than 1 billion transactions each week.
Malanga said revenue across subnet operators had risen from almost nothing over an 18-month period to a recent estimate exceeding $32 million. He cited that figure, together with customer agreements, as evidence investors should examine when assessing whether Bittensor network activity reflects practical use rather than speculation.
The supplied Yuma background also identified Innerworks, a London-based bot-detection company, as a business incorporating Bittensor competition results into its products. According to that account, Innerworks improved its detection rate from 73% to 99.5% in less than a year.
How Bittensor’s 128 AI subnets work
Malanga said Bittensor currently supports 128 markets, known as subnets, covering tasks that include mathematical proofs, machine learning, prediction and resource optimization. Each subnet defines a specific task. Miners compete to produce solutions, while validators evaluate the quality and usefulness of that work.
Malanga compared the process with Bitcoin mining, where participants compete to find a valid nonce, but emphasized that Bittensor is designed to reward multiple forms of machine intelligence.
“Bitcoin rewards one type of work, while Bittensor can reward many forms of machine intelligence.”
Under Malanga’s explanation, a task resembling Bitcoin’s proof-of-work process could operate inside a single Bittensor subnet. The network separates the creation of intelligence from the assessment of its quality, giving miners and validators distinct roles.
Earlier coverage published on March 25 documented growth in subnet staking, with the value of $TAO staked across subnets exceeding $620 million. That report also recorded an increase in the number of subnets from approximately 80 to more than 120 during the period examined.
Validator scoring determines $TAO reward distribution
Each subnet establishes its intelligence task and sets the criteria validators use to assess competing solutions, according to Malanga. Independent validators score the submissions, and miners receive rewards based on the utility assigned to their outputs.
Bittensor combines those assessments through Yuma Consensus. Malanga described Yuma Consensus as a mechanism intended to limit collusion while enabling agreement, including in situations where parts of an evaluation are subjective.
Miners whose solutions receive higher utility scores obtain a larger share of newly issued tokens, according to Malanga. This makes validator scoring central to the way Bittensor distributes rewards among competing participants.
Malanga separately identified external customer demand as the key measure of whether those rewards translate into lasting commercial value.
“Lasting commercial value is ultimately determined by external demand for the intelligence.”
He cited Lium for GPU computing, Chutes for AI inference, Leadpoet for AI sales intelligence, Score for computer vision and RedTeam for cybersecurity as examples of services available through Bittensor subnets. When asked whether participants could influence rewards without creating durable value, Malanga pointed to defined tasks, independent evaluations and Yuma Consensus. His explanation of commercial value rested on whether outside customers continue to demand the intelligence produced.
Risks for $TAO investors and Bittensor adoption
Earlier Bittensor coverage documented U.S. equity-market exposure to the network through a Nasdaq-listed treasury company. In August 2025, $TAO Synergies disclosed treasury holdings of 42,111 $TAO, consisting of tokens acquired and generated through staking.
The company had purchased $10 million worth of $TAO in July 2025 as part of a strategy focused on Bittensor, according to that report. The disclosure provides historical context for investors seeking exposure to Bittensor through a listed company rather than through direct token ownership.
The supplied background said Yuma, a DCG-backed company, builds and invests in infrastructure for specialized open-source AI on Bittensor. Yuma also described itself as operating the network’s largest owned-hardware validator, with more than $200 million in $TAO and subnet tokens staked.
Malanga identified several risks to that investment position: individual subnets may fail to develop commercially viable businesses, early-stage digital assets can remain volatile, and Bittensor faces competition from other AI platforms, including frontier laboratories. Although he said demand for AI is clear, he noted that it remains uncertain which platforms will ultimately achieve scale.
Yuma’s approach is to use diversified subnet investment vehicles to reduce exposure to the volatility of individual projects while maintaining investment across the wider Bittensor ecosystem.
Why This Matters
Bittensor’s development is increasingly being assessed on two connected fronts: the growth of its token economy and the ability of subnet operators to deliver services that businesses will pay for. Commercial deployments involving Avia carwash locations, financial institutions, transaction protection and bot detection offer concrete examples of how Bittensor competitions may feed into products outside cryptocurrency trading.
The network’s reward model still depends on validator assessments and token issuance, but the examples cited by Malanga place external demand at the center of the long-term adoption question. Revenue growth, customer agreements and continued use of subnet services therefore remain important indicators as investors weigh Bittensor’s potential against subnet failure, asset volatility and competition across the AI industry.
Frequently Asked Questions
What are Bittensor subnets?
Bittensor subnets are specialized markets within the network that focus on tasks such as mathematical proofs, machine learning, prediction, resource optimization, computer vision and cybersecurity. Miners produce solutions and validators assess their quality.
How are Bittensor miners rewarded?
Validators score miners’ work according to each subnet’s criteria. Bittensor aggregates those assessments through Yuma Consensus, and miners whose outputs receive higher utility assessments receive a larger share of newly issued tokens.
What are the main risks associated with Bittensor?
The risks identified by Malanga include individual subnets failing to become commercial businesses, volatility in early-stage assets and competition from other AI platforms, including frontier laboratories. It also remains uncertain which AI platforms will ultimately achieve significant scale.