Mentioned in This Article
1. Anza puts numbers on the 2027 Solana roadmap
Anza CEO Brennan Watt used the Solana Breakpoint 2026 conference this week to present a 2027 Solana roadmap that aims to roughly double the network's compute capacity to about 500 million compute units per second. The plan also targets an 80% cut in the compute needed for token-account transactions, lower storage costs, a new block-propagation protocol called Rotor and a transaction-ordering effort called Constellation.
Anza's stated goal is to make on-chain trading fast and cheap enough to compete with centralized trading platforms. The targets are engineering objectives rather than shipped features, and the company has not yet attached activation dates to most of them.
The presentation comes a month after Solana activated Alpenglow, a consensus overhaul also designed by Anza, and a few months after the network raised its per-block compute limit. Together, those steps form the foundation the new roadmap builds on.
2. Who is Anza?
Anza is a Solana engineering firm that spun out of Solana Labs in January 2024. Its main product is Agave, the validator client that a large share of the network runs. A validator client is the software that operators use to process transactions, produce blocks and take part in consensus, so changes to Agave shape how the entire network behaves.
That position gives Anza's roadmap unusual weight. While any Solana upgrade still has to move through the community proposal process and be adopted by validators, the team maintaining the most widely used client has a major say in what gets built and when.
Solana's validator ecosystem includes more than one client, a setup intended to reduce the risk that a single software bug could halt the network. Even so, coordinated upgrades still depend on the major client teams aligning on design choices and timelines, and Anza is central to that coordination.
Watt's presentation therefore reads less like a wish list and more like a statement of where much of Solana's core engineering effort will go over the coming year.
3. Doubling compute to 500 million units per second
The headline target is a network capable of handling about 500 million compute units every second, roughly twice its current capacity. Compute units are Solana's measure of how much processing work a transaction requires. Every instruction a program executes consumes some of them, and the network caps how many can be used per block so that validators are not overwhelmed.
Raising total capacity means more transactions, or more complex transactions, can fit into the same span of time. For applications such as trading venues, where order placement, cancellation and settlement all happen on-chain, that headroom translates directly into how much activity the network can absorb during busy periods before fees rise or transactions are delayed.
The compute target is paired with an efficiency goal. Anza wants to cut the compute units consumed by token-account transactions by roughly 80%. Token accounts hold balances of tokens issued on Solana, including stablecoins, so operations involving them sit at the heart of almost every trade, payment and transfer. Making each of those operations cheaper in compute terms frees up capacity even before the overall ceiling rises.
Compute also feeds into what users pay. When demand for block space outstrips supply, transactions compete through priority fees, and the busiest moments in the market tend to be exactly when fees spike. A higher ceiling and leaner token operations both ease that competition, which is why Anza presents them as two halves of the same effort rather than separate projects.
4. Cheaper storage under SIMD-0437
Storage is the third major target. On Solana, keeping data on-chain carries a cost, because every account must hold a minimum balance to remain stored by validators. That requirement protects the network from being flooded with data, but it also makes creating accounts more expensive for users and developers.
Watt said storage costs should fall about sevenfold further following recent upgrades, as the remaining steps of the SIMD-0437 proposal take effect, with further improvements expected in the weeks after his talk. SIMD stands for Solana Improvement Document, the formal process through which changes to the protocol are proposed and debated.
For developers, the deposit tied to each account is a design constraint that shapes how applications are built. Some teams consolidate data into fewer accounts or close unused ones to recover deposits. A meaningful reduction in that cost lowers the barrier to building products that need many accounts, from token launches to applications serving large numbers of new users.
Lower storage costs matter most for applications that open large numbers of accounts, such as token issuers, trading platforms and consumer apps onboarding new users. The storage change is also the part of the roadmap closest to delivery, which makes it an early test of how quickly Anza's plans move from presentation to mainnet.
5. Rotor and Constellation: how blocks travel and how trades are ordered
Two protocol projects round out the performance agenda. The first, Rotor, is planned as the successor to Turbine, the system Solana currently uses to spread newly produced blocks across the network. Turbine breaks blocks into small pieces and relays them through layers of validators so that no single node has to send the full block to everyone. Replacing it is meant to improve how quickly and reliably block data reaches the rest of the network.
The second, Constellation, focuses on how transactions are ordered. Ordering is a sensitive issue on any blockchain used for trading, because the sequence in which orders land can determine who gets filled at what price. Clearer and fairer ordering rules are a prerequisite for the kind of high-volume, professional trading that Anza hopes to attract.
Neither Rotor nor Constellation is live. Both are listed as planned work, and they sit alongside the compute and storage targets as part of the broader 2027 agenda.
6. What has Solana already shipped?
The new roadmap follows two significant upgrades in 2026. The larger of the two was Alpenglow, a consensus upgrade designed by Anza that went live on Solana mainnet on October 21, 2026. It was built to shorten transaction finality, the point at which a transaction can no longer be reversed, from about 12.8 seconds to roughly 150 milliseconds.
Finality is particularly important for trading. Until a transaction is final, a counterparty cannot be fully certain that a fill or a transfer will stand, so shorter finality reduces the window of risk for market makers and arbitrage traders who move capital quickly between venues.
Earlier, in July 2026, the network raised its per-block compute-unit limit to 100 million. That change allowed each block to carry more work and was an early step in the steady expansion of Solana's capacity.
Proposals to remove compute-unit limits altogether now that Alpenglow is active are also under discussion in the Solana developer community. Against that backdrop, Anza's push for more capacity looks like the natural continuation of a direction the network has already taken, rather than an abrupt change of course.
7. Why trading sits at the center of the plan
Every major element of the roadmap points toward the same destination: on-chain trading that can stand up to centralized platforms on speed and cost. Centralized exchanges match orders on private servers and settle balances in internal databases, which lets them process enormous volumes with very low latency. Their trade-off is custody, since users must hand over their assets and trust the operator.
Decentralized venues remove that custody risk but have historically struggled with throughput, fees and the fairness of transaction ordering. More compute capacity addresses throughput. Cheaper token-account operations and lower storage costs address fees. Rotor addresses how quickly block data spreads, and Constellation addresses ordering. Alpenglow's faster finality addresses the time it takes for a trade to become irreversible.
Taken together, the plan is an attempt to close each of the gaps that keep large traders on centralized platforms. Whether it succeeds depends on execution and on whether traders actually move once the technical barriers fall.
8. The gap between roadmap and mainnet
For all its ambition, most of the roadmap remains unfinished. Rotor, Constellation, the 500 million compute-unit target and the deeper efficiency gains are plans, not live features, and there is a clear distance between what was presented on stage and what is running on mainnet today.
Blockchain upgrades of this size typically pass through public proposals, test networks and staged rollouts before they reach production, and validators must adopt the new software. Each of those steps can introduce delays, especially for changes that touch consensus, networking or fee mechanics.
Solana's track record this year offers some reassurance, with Alpenglow and the 100 million compute-unit limit both now active. But the network's history also includes periods of instability, which is why stability under heavier load is likely to be the measure that matters most to users and validators.
9. What to watch next
Three checkpoints will show whether the roadmap is on track. The first is whether compute-unit limits are actually lifted as proposed. The second is whether larger blocks arrive without stability problems, a crucial test for a network that wants to host professional trading. The third is whether the storage cost reductions land on schedule, with the next round expected in the weeks following Watt's presentation.
Beyond those near-term markers, progress on Rotor and Constellation will indicate how quickly the deeper protocol changes are moving. Neither has been given an activation window, so updates through the Solana Improvement Document process will be the main signal to follow.
For traders, developers and validators, the 2027 Solana roadmap sets a clear ambition: twice the capacity, far leaner token transactions and cheaper storage, all aimed at making the network a credible home for trading that today happens mostly on centralized platforms.
