Solana has activated SIMD-0286 on mainnet, increasing the maximum computational workload permitted within each block from 60 million to 100 million compute units. The 66% capacity increase took effect on July 29, 2026, at the beginning of epoch 1009, following earlier activations on the network’s testnet and devnet environments.
Compute units measure the processing work performed by transactions and program instructions. The block-level limit determines how much aggregate work a leader can include during a slot. Raising that ceiling gives Solana additional room to process activity during periods of elevated demand, although it does not guarantee a proportional 66% increase in the number of completed transactions because individual operations consume different amounts of compute.
Additional Capacity Targets Demand Spikes
The previous 60 million-unit limit had been tested repeatedly by live network traffic. Between its activation on July 22, 2025, and the latest upgrade, 11.2% of blocks consumed at least 56 million compute units. Approximately one in nine blocks operated close to the former ceiling, with pressure tending to intensify during volatile market periods when traders competed more aggressively for blockspace.
SIMD-0286 is intended to provide additional headroom during those bursts. A larger block budget may allow more swaps, payments and other transactions to land without competing against an almost-full block. The upgrade expands Solana’s theoretical throughput without changing how developers construct applications or users submit transactions. Solana classifies the activation as non-breaking and says no indexing changes are required.
The adjustment affects only the network’s total block compute allowance. The maximum workload that can write to one account remains fixed at 12 million compute units, while the maximum increase in account-data size per block remains 100 MB. A heavily used application or account can therefore continue reaching its existing local constraint even when unused capacity remains elsewhere in the block.
Holding the account-level limit steady shifts the additional capacity toward parallel activity. Under the previous configuration, one account could consume as much as 20% of a block’s compute allowance. Under the new ceiling, the same 12 million-unit workload represents 12%. A congested program now occupies a smaller share of total block capacity, leaving more room for unrelated applications to execute alongside it.
Infrastructure Must Absorb Larger Blocks
The capacity increase also creates a larger potential workload for validators and supporting infrastructure. Blocks containing more computation can take longer to execute and replay, while nodes that fall behind may require additional time to catch up. RPC providers, indexers, exchanges and validator operators must ensure their systems can handle sustained 100 million-unit blocks, even though the upgrade does not modify transaction or indexing formats.
Solana linked the rollout to improvements in validator networking, particularly the adoption of XDP, a kernel-bypass system designed to accelerate packet processing. XDP is available in Agave 4.0.0 and later, enabled by default in Firedancer and was active across more than 70% of mainnet stake before the upgrade. Core engineers determined that network propagation had sufficient headroom to support the larger block limit.
SIMD-0286 continues a staged expansion of Solana’s compute capacity. The network previously moved from 50 million to 60 million units through SIMD-0256 in July 2025. Developers considered an intermediate increase to 80 million but selected the full 100 million-unit target after assessing improvements in validator-client performance. Solana has now doubled its block compute ceiling from the 50 million-unit baseline used before the two upgrades.
The activation is an infrastructure expansion rather than evidence that network demand has immediately increased. Its practical value will depend on how frequently applications consume the additional blockspace and whether validators can maintain reliable propagation and replay performance during sustained peaks. Transaction inclusion, fee competition and infrastructure stability will provide the clearest indication of whether the added headroom produces durable throughput gains.