Bittensor adoption grows as AI subnets find paying customers, Yuma CRO says

Bittensor subnet operator revenue has grown from near zero to an estimated $32 million-plus over 18 months as companies have secured commercial implementation deals, according to Yuma chief revenue officer Evan Malanga.

Evan Malanga, chief revenue officer at Bittensor infrastructure company Yuma, told crypto.news that customer deals, operator revenue and subnet token purchases offer evidence of growing use across the network.

In his account, the strongest examples include Score, which has planned carwash monitoring deployments for Avia across thousands of locations, and RedTeam, which serves multiple Fortune 500 financial institutions protecting more than 1 billion weekly transactions.

Malanga presented the deals as examples of demand for services produced through Bittensor’s AI competitions. He also pointed to an increasing number of operators buying their subnet tokens on the open market, saying Bittensor’s design connects demand for those assets with demand for its native token, TAO.

Bittensor adoption reaches carwash monitoring and cybersecurity

For Malanga, commercial deployments provide a way to examine Bittensor beyond token trading. His examples cover computer vision and cybersecurity, two areas where subnet operators have secured implementation agreements with businesses.

Score’s planned rollout involves monitoring Avia carwash locations, according to the executive. RedTeam’s work concerns financial institutions and transaction protection, with Malanga putting the volume covered by its customers at more than 1 billion transactions each week.

Across subnet operators, he said revenue had risen from almost nothing over an 18-month period to a recent estimate exceeding $32 million. He cited the revenue estimate alongside customer agreements when asked what evidence investors should examine to distinguish network use from speculation.

The supplied Yuma background also described Innerworks, a London bot-detection company, as an example of a business using Bittensor competitions in its products. According to that account, Innerworks increased its detection rate from 73% to 99.5% in less than a year.

Yuma described the competitions as task-specific contests with public scoring, where teams receive tokens for producing the best-performing AI work. In the company’s account, businesses running many of the competitions incorporate winning outputs into their own products.

TAO rewards work across 128 AI markets

Explaining the network’s structure, Malanga said Bittensor supports 128 markets, known as subnets, covering tasks such as mathematical proofs, machine learning, prediction and resource optimization.

Within each subnet, miners compete to answer a defined task, while validators assess the quality of their work, he said. Malanga compared the competitive process with Bitcoin mining, where participants compete to find a valid nonce, but stressed that Bittensor can reward several kinds of machine intelligence.

“Bitcoin rewards one type of work, while Bittensor can reward many forms of machine intelligence.”

Under his explanation, a task resembling Bitcoin’s proof-of-work process could conceivably run inside a single Bittensor subnet. Bittensor separates the production of intelligence from the evaluation of its quality, he said, allowing miners and validators to perform different roles.

Earlier coverage published on March 25 documented growth in subnet staking, with the value of TAO staked across subnets exceeding $620 million. The report also recorded an increase in the subnet count from about 80 to more than 120 over the period it examined.

For token demand, Malanga pointed to operators purchasing their own subnet assets. According to him, the protocol directly connects demand for subnet tokens with TAO demand, while demand for the resulting AI services comes from customers seeking the work those markets produce.

Validator scoring determines how miners receive rewards

On reward allocation, Malanga said each subnet defines an intelligence task and the criteria validators use to assess competing solutions. Independent validators score the work, and miners receive rewards according to the usefulness assigned to their outputs.

Bittensor aggregates the assessments through Yuma Consensus, which he described as a mechanism designed to resist collusion while allowing agreement, including when parts of an evaluation are subjective.

According to the executive, miners whose solutions receive higher utility assessments collect a larger share of newly issued tokens. His description places validator scoring at the center of how the network distributes rewards among competing participants.

For lasting commercial value, however, Malanga pointed to demand outside the reward system.

“Lasting commercial value is ultimately determined by external demand for the intelligence.”

He named Lium for GPU computing, Chutes for AI inference, Leadpoet for AI sales intelligence, Score for computer vision and RedTeam for cybersecurity as examples of the services available through subnets.

When asked about participants influencing rewards without producing lasting value, Malanga cited the combination of defined tasks, independent evaluations and Yuma Consensus. His explanation of commercial value separately rested on whether outside customers demand the intelligence produced.

U.S.-listed treasury exposure carries token and adoption risks

For U.S. equity investors, earlier Bittensor coverage documented a Nasdaq-listed company holding TAO as a treasury asset. In August 2025, TAO Synergies disclosed 42,111 TAO treasury holdings, comprising tokens acquired and generated through staking.

The company had purchased $10 million worth of TAO in July 2025 as part of its strategy to concentrate on Bittensor, according to that report. The disclosure provides historical context for American stock-market exposure to the network through a listed treasury company.

Within Yuma’s own business, the supplied background said the 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.

Asked about risks to that investment position, Malanga identified individual subnets failing to turn their work into commercial businesses, early-stage asset volatility and competition from other AI platforms, including frontier laboratories.

Although demand for AI is clear in his view, he said the platforms that will ultimately achieve scale remain uncertain. Malanga described diversified subnet investment vehicles as Yuma’s approach to reducing exposure to the volatility of individual projects while investing across the Bittensor ecosystem.