Skip to content
Advertisement
Solana targets doubling compute capacity, cutting costs by 80% by 2027

Solana targets doubling compute capacity, cutting costs by 80% by 2027

Solana Foundation official brand assets (solana.com/branding)

Anza's 2027 roadmap aims for 500 million compute units per second, cheaper token transactions, and on-chain vector processing

Solana wants to double its workload capacity while charging users less to use it.

At the Solana Breakpoint 2026 event in London, Anza, the core development firm behind the network, laid out a technical roadmap for 2027. The headline goals: double the network’s compute capacity and cut the compute needed for token-account transactions by 80%.

What Anza is promising

Anza CEO Brennan Watt presented the plan. The biggest number is throughput: the network currently benchmarks at 250 million compute units per second, and the target is 500 million by the end of 2027.

Doubling that ceiling means the chain could, in theory, handle twice the computational workload.

The second goal targets the most common type of activity. Anza aims for an 80% reduction in the compute units needed for transactions involving token accounts.

Token accounts are where balances of specific tokens live on Solana. If you’ve swapped a memecoin, received a stablecoin, or moved almost anything that isn’t SOL itself, you’ve touched one.

Advertisement

Making those operations cheaper in compute terms has a double effect. Each transaction costs less to process, and the freed-up capacity can be used by other activity on the network.

The roadmap also includes on-chain vector processing. This is meant to support advanced computational models running directly on the blockchain.

Storage rent, more accounts, and faster finality

A proposal called SIMD-0437 targets storage rent, the fee for keeping data stored on the network, with an overall reduction goal of roughly 90%.

The first step brings the cost down from 6,960 to 5,080 lamports per byte. Lamports are the tiny units SOL is broken into, so this is a per-byte price tag getting trimmed.

That initial cut is only a partial move toward the ~90% target. The remaining reductions would need to follow in later steps.

Anza also plans to raise the number of accounts a single transaction can touch, from 64 to 96.

This sits on top of an upgrade the network has already shipped. The Alpenglow upgrade, implemented on October 21, 2026, cut transaction finality from about 12.8 seconds to a target of 150 milliseconds.

What this means for Solana

Anza’s approach leans on efficiency rather than brute force. Cutting the compute needed for token-account work by 80% frees space without needing every validator to buy a bigger machine. Doubling headline capacity then adds more room on top.

That’s the key caveat. These are targets, and the main capacity goal sits at the end of 2027.

Solana does have a recent delivery to point to. The Alpenglow upgrade went live and reshaped finality, which gives the 2027 plan more credibility than a pure wish list.

What to watch next: progress on SIMD-0437’s later rent reductions, the rollout timeline for the 96-account limit, and measured throughput as the network moves from 250 million toward 500 million compute units per second.

Disclosure: This article was edited by Estefano Gomez. For more information on how we create and review content, see our Editorial Policy.