The growing volume of autonomous AI agents poses a structural challenge to block space capacity on shared Layer 1 blockchains, according to Bart Smith, CEO of Avalanche Treasury Company (AVAT). Smith contends that networks unable to isolate workloads will face increasing strain as AI-driven transactions multiply.
By mid-February 2026, Avalanche had registered over 1,600 AI agents operating under the ERC-8004 standard. These agents actively trade and execute on-chain tasks, transacting continuously without regard for fluctuating gas prices or network congestion, compounding transaction volumes.
Smith, speaking on the July 9 episode of the Layer One podcast, highlighted that blockchain capacity is finite. Shared environments like Ethereum and Solana, where all applications compete for the same block space, will experience heightened contention. This dynamic results in higher fees and slower confirmations for all users when AI trading bots flood the network during volatile periods.
Avalanche's proposed solution is its subnet model. This architecture enables organizations to launch purpose-built Layer 1 blockchains that operate independently. An AI-heavy trading subnet cannot consume throughput intended for a tokenized real estate subnet; workloads remain distinct.
AVAT, which went public on Nasdaq in June 2026 through a SPAC merger valued at approximately $675 million, demonstrates institutional conviction in Avalanche's strategy. The company has set an AVAX treasury target exceeding $1 billion.
Avalanche currently supports around 80 live purpose-built Layer 1 blockchains, with projections to expand toward 200 custom institutional chains. These subnets are tailored for specific industry applications, ranging from real-world asset tokenization to compliance-intensive financial products.
Shared Layer 1s offer broad network effects and established security models, but Smith argues they face a fundamental structural limitation. Even scaling solutions like Ethereum rollups still share security and settlement with the mainnet, meaning network-wide congestion can still impact them indirectly.
Institutions, already cautious about integrating with unpredictable DeFi protocols on shared infrastructure, will become more hesitant to share block space with thousands of autonomous agents they do not directly control. The AI agent explosion accelerates this need for dedicated, isolated environments.
Smith anticipates traditional financial markets will transition to 24/5 trading by mid-2027. This shift will require blockchain infrastructure capable of handling continuous, high-volume operations without cross-application interference. Avalanche's subnet model positions it as a competitive advantage for this institutional demand.
