0x18cf48fd18cf…18cf48fa
Solana Foundation sets AI agent payment and benchmark roadmap
Solana Foundation AI initiatives head Rishin Sharma detailed a 12-month plan for benchmarks, inference access and card-to-stablecoin funding after x402 topped 60 million transactions in two months.

Outputs
On November 17, 2026, Rishin Sharma presented “The Next 12 Months of AI on Solana” at Breakpoint 2026 in London.
The roadmap covers financial benchmarks for AI models, model inference and a card-to-stablecoin onramp for agents.
Solana.ai had launched one week earlier and listed more than 120 AI-product teams, with payments the leading use case.
x402 processed more than 60 million transactions in two months and more than 400 million payments over its lifetime, alongside $100 million in volume.
Agents generated 76% of OpenRouter token volume by October 2026, up from 8% a year earlier.
At Breakpoint 2026 in London on November 17, 2026, Solana Foundation head of AI initiatives Rishin Sharma outlined financial benchmarks, model inference and a card-to-stablecoin onramp for AI agents, citing research that the x402 payment standard had processed more than 60 million machine-to-machine transactions in two months.
The foundation’s operating thesis is that AI agents will become central users of onchain infrastructure and that its next wave of users will be software. Sharma’s presentation, “The Next 12 Months of AI on Solana,” put that shift at the center of the foundation’s roadmap.
The launch of Solana.ai came one week before Sharma’s presentation. At the time, the hub listed more than 120 teams building AI-based products on Solana, with payments the leading use case. That concentration puts agent payment demand at the center of the foundation’s user-growth strategy.
What does this roadmap change for AI agents?
For an AI agent operator, the practical effect is a proposed stack that covers evaluation, usage costs and funding. Sharma’s roadmap has three components:
- Financial benchmarks for AI models. Sharma did not identify the benchmark criteria or participating models.
- Model inference access. Inference is the process of running a trained model to produce an answer.
- A card-to-stablecoin onramp. The proposed route is designed to fund AI agents.
The presentation did not name a card issuer, supported stablecoins or implementation dates. Those omissions leave open basic commercial and settlement questions for operators: which assets agents can hold, who provides the card connection and when the components become available.
According to the research cited by Sharma, agents accounted for 76% of OpenRouter token volume by October 2026, compared with 8% a year earlier. That is a 68-percentage-point increase and a near-tenfold rise in share.
OpenRouter routes requests across many AI models, so its traffic offers a rough read on who is using them. It does not, by itself, establish Solana-specific demand.
Sharma said Solana’s payment channels can handle 1 million payments per second. The cited x402 research recorded more than 60 million transactions over two months, more than 400 million payments across the standard’s lifetime and $100 million in volume. The capacity and activity figures measure different things, but together they give the roadmap an existing payment base.
Pay.sh provides a more concrete example, integrating more than 100 APIs that agents can pay to access. For developers, the arrangement makes charging part of the API interaction available to autonomous software.
Sharma also cited partnerships with Google, Alibaba and ByteDance focused on pay-per-request access to AI models. Under that model, an agent pays for exactly what it uses, one call at a time, instead of carrying a monthly subscription.
Inference is the process of running a trained AI model to produce an answer, and it incurs a cost each time someone asks a question. Metered access aligns the payment with the operation that consumes resources.
Sharma’s presentation also sat within a broader Breakpoint theme that the conference called a “token supercycle.” The phrase connected AI functionality with stablecoins, tokenized assets and programmable capital on Solana. It was a conference framing, not a disclosed investment forecast or implementation timetable for the foundation’s plan.
Over the 12-month window Sharma identified, the foundation will need to turn benchmarks, inference access and card funding into deployable services. Without a named issuer, supported stablecoins or launch dates, the roadmap still describes a market structure to build rather than a scheduled rollout.
via Crypto Briefing (Source)
More from Nathan Brooks
Show full bio
Market editor covering money, funding and market shifts around crypto and web3 industry at Mempool Brief.
1,512 articles