---

The CEO of Avalanche Treasury Company told the Layer One podcast on July 9, 2026 that blockchain capacity can no longer be treated as a limitless resource. "There's not enough block space. And block space is not infinite anymore," said Bart Smith, whose firm holds AVAX as a corporate treasury asset and targets a position exceeding $1 billion. Avalanche Treasury Company, trading on Nasdaq under the ticker AVAT, went public in June 2026 via a SPAC merger valued at approximately $675 million. The Block reported his remarks on September 17. Smith's core claim: autonomous AI agents are multiplying fast enough to crowd out ordinary users on networks that were not designed for workload separation.

The argument is not purely theoretical. AI agents now account for roughly 19% of all on-chain transaction volume globally, according to data cited by Startup Fortune, driven by an estimated 1.4 million active deployments across Ethereum, Solana, and Base. On Solana, agents represent nearly 35% of non-voting transaction throughput during peak hours. In Q1 2026, AI-related transaction fees across major chains surpassed $450 million. A more dramatic illustration came in February 2026, when 15,000 AI agents attempted to exit the same liquidity pool at the same moment, liquidating $400 million in leveraged positions within three seconds. That event, widely referred to as the "February Wick," showed how machine-speed coordination can impose systemic stress on a shared chain.

Avalanche recorded 235.6 million transactions in Q2 2026, its seventh consecutive quarter of growth, with a daily average of roughly 2.5 million and a peak of 3.6 million in early May, according to a Nansen report. Stablecoin transfer volume reached $84.4 billion for the quarter. The network's median transaction fee fell to $0.000014, a 99.6% year-over-year decrease. Avalanche's Real-World Asset market reached $1.65 billion at the close of Q2 2026, up 27% quarter-on-quarter, and the network has maintained sub-one-second block times since April 2026. Both figures are directly relevant to the institutional capacity argument Smith is advancing. By February 2026, more than 1,600 AI agents had registered on the network under the ERC-8004 standard, a registration protocol that assigns on-chain identity to autonomous agents, though that figure is now roughly seven months old and requires verification against Avalanche Explorer or official Ava Labs data before publication.

Smith's proposed solution is Avalanche's subnet architecture. Each subnet is its own independent blockchain, purpose-built for a specific use case, so that high-frequency AI trading activity cannot crowd out, say, tokenized real estate settlement. "An AI-heavy trading subnet doesn't cannibalize throughput on a tokenized real estate subnet," Smith said. Avalanche currently supports around 80 live institutional blockchains, with published projections pointing toward 524 active Avalanche L1s and 200 dedicated institutional chains. Smith also contended that traditional financial markets are heading toward 24-hour weekday trading by mid-2027, a shift that existing settlement infrastructure cannot support but purpose-built chains potentially could. That timeline is his projection, not confirmed policy, and should be read as a forecast rather than a scheduled industry change. The direction of travel is, however, independently supported: the NYSE is building a 24/7 blockchain-based platform for tokenized securities trading and the London Stock Exchange has announced a 24-hour trading venue, both discussed in greater detail below.

The argument carries an obvious commercial dimension. AVAT's treasury is denominated in AVAX, giving Smith a direct financial incentive tied to Avalanche network growth and adoption, regardless of Ava Labs's position. Ava Labs, the separate entity that developed the Avalanche protocol, also benefits when developers and institutions choose subnet infrastructure over competing L1s, aligning its interests with AVAT's even though the two organisations are distinct. No public response from Ethereum Foundation or Solana Labs teams was available at the time of publication, and independent benchmarking of how those networks would handle equivalent AI agent loads was not located. AVAX itself is trading at approximately $7.51, roughly 95% below its all-time high of $144.96, with a market cap near $3.2 billion and 24-hour trading volume around $355 million. The bullish infrastructure narrative Smith is promoting should be weighed against that market context.

For users in South Asia and Sub-Saharan Africa, the blockspace question carries practical weight. Sub-Saharan Africa received over $205 billion in on-chain value between mid-2024 and mid-2025, a 52% increase year over year. Nigeria processed more than $30 billion in DeFi value in 2024, and Kenya ranks fifth globally for transactional stablecoin use. Micro-transactions, cross-border remittances, and informal commerce payments in the region typically fall between $1 and $10, a range where fee volatility matters enormously. Traditional remittance corridors charge 6 to 8%. Blockchain alternatives currently offer fees of 1.5 to 2.5%, but congestion-driven fee spikes on shared chains can rapidly erase that cost advantage. If AI agent activity pushes gas costs above $0.10 on high-traffic networks, the use cases driving African adoption become economically unviable. This is an analytical inference based on fee trend data, not a finding from a direct study of AI agent impact on African users specifically. The regulatory environment adds further complexity: Nigeria's Investment and Securities Act 2025 formally recognised digital assets as securities, and Kenya has implemented a 10% excise duty on crypto transaction fees. Both measures directly affect the cost structure and legal feasibility of purpose-built chains in the region. Human review is required to confirm the current status of both legislative instruments before publication.

In India, early deployments of agentic payment systems are already live. Razorpay and NPCI announced a pilot in February 2026 using Anthropic's Claude to process consumer payments on platforms including Zomato and Swiggy. Gnani, another Indian firm, completes UPI payments inside voice calls across more than 22 languages, a capability it announced at the India AI Impact Summit on February 20, 2026. These applications generate the kind of high-frequency, low-value on-chain activity that Smith warns will compete for space on congested shared chains. India's UPI-native user base, accustomed to near-zero transaction costs, would be among the first to notice if fees rise. A concrete illustration of the mixed-workload risk is already visible on the ground: Arya.ag, an Indian agri-fintech, uses Avalanche to record grain ownership across farmers, warehouses, insurers, and lenders. AI agents performing continuous inventory and collateral checks on the same chain as retail DeFi users represent precisely the scenario Smith warns against. The Reserve Bank of India has taken a broadly permissive stance on blockchain infrastructure while remaining cautious on crypto-native assets, a distinction that is material to whether Indian financial institutions could realistically adopt purpose-built subnets. Human review is required to confirm the RBI's current position before publication. South Korea offers live examples of the subnet model in operation. Hanwha Investment and Securities built a tokenized securities platform on Avalanche, and NHN KCP, the backbone of South Korean e-commerce with more than $38 billion processed in 2025, launched South Korea's first payment-dedicated Avalanche L1 blockchain.

The institutional infrastructure Smith points to as validation is real. T. Rowe Price (approximately $1.6 trillion in assets under management), WisdomTree (more than $110 billion in ETFs), Wellington Management (approximately $1.3 trillion in assets under management), and Cumberland are all part of the Spruce Evergreen subnet cohort that moved from testnet to production in April 2026. In a related deployment, JPMorgan's Onyx ran a tokenized portfolio rebalancing pilot on a permissioned Avalanche subnet alongside WisdomTree and Apollo, further substantiating the institutional uptake case. The NYSE is developing a 24/7 blockchain-based platform for tokenized securities trading, and the London Stock Exchange has announced LSE 24, a 24-hour venue planned for 2027 that will list xStocks. Whether those developments translate into a fee crisis on shared chains, or whether Avalanche's subnet model is the right response, remains an open question. But the transaction volume data suggests the pressure Smith describes is already building.