Zimbuck – Your Market. Your Community.
Solana Blockchain Unveils 2027 Roadmap to Double Compute Capacity and Run AI Models Fully On-Chain

Solana Blockchain Unveils 2027 Roadmap to Double Compute Capacity and Run AI Models Fully On-Chain

5 min de leitura
0
Comentários
2
Visualizações

Solana's core developers have set out what the blockchain should look like in 2027, and the headline target is raw capacity: double the network's compute from about 250 million to 500 million compute units per second, while making everyday token transactions far cheaper to process.

 

The roadmap was presented at the Breakpoint conference in London by the team behind the network's main client software. It is a plan, not a release — no firm date was given for each change — but it shows exactly where one of the busiest blockchains in the world thinks its bottlenecks are. Follow blockchain coverage daily on the Zimbuck News page at https://zimbuck.com/pages/zimbucknews .

 

The four pillars of the 2027 plan

 

The roadmap is organised around four areas:

 

- Performance — double total compute capacity, and cut the compute units used by token-account transactions by about 80%

- Expressivity — new programming features, including vector processing so AI models can run fully on-chain

- Market structure — faster, fairer block distribution and clearer transaction ordering rules

- Resilience — a simpler, tougher protocol after this year's big consensus upgrade

 

Building on Alpenglow

 

The starting point matters. In October, the network activated its Alpenglow upgrade, which cut transaction finality from about 12.8 seconds to roughly 150 milliseconds. The 2027 work builds directly on that base: a follow-on set of consensus improvements pipelines leader handoffs more efficiently and adds geographic awareness inside the protocol, while a new system called Rotor is intended to replace Turbine, the mechanism that spreads blocks across validators. A companion proposal, Constellation, sets clearer rules for how transactions are ordered — a detail traders care about because ordering decides who gets filled first when the market moves fast.

 

Why on-chain AI is in a blockchain roadmap

 

The most eye-catching item is vector processing in the virtual machine. In plain terms, the developers want the chain itself to handle the maths that AI models need, so a model could execute entirely on-chain instead of calling out to a server. Early targets named for this work are on-chain options pricing and more sophisticated margin models — financial calculations first, which fits where Solana's activity already concentrates. Alongside it, the maximum number of accounts a single transaction can touch rises from 64 to 96, and transactions will read account data directly instead of copying it, reducing the data each transaction carries.

 

Traders and builders are debating what a faster, AI-capable base layer means for the apps on top of it in the Zimbuck Market News group at https://zimbuck.com/groups/zimbuckmarketnews .

 

Cheaper storage is coming sooner

 

One change lands well before 2027. Storing data on the network is set to get roughly seven times cheaper within weeks, as the final rent steps cut the rate from 5,080 to 696 lamports per byte. Cheaper storage lowers the cost of running apps that keep a lot of state on-chain — market data, game state, tokenised asset records — and it is the kind of unglamorous change that quietly decides whether new products are viable.

 

Our recent Blockchain & Web3 coverage on Zimbuck follows how tokenised equities have been growing on this same network, and it is the right companion read for this roadmap: https://zimbuck.com/blogs/5574/Solana-Tokenized-Stocks-Pass-1-22-Billion-as-Blockchain-Equities

 

What to watch next

 

Three tests will show whether this roadmap is delivery or presentation. First, whether the compute doubling arrives as one upgrade or in phases, and what it does to validator hardware requirements — more capacity is only decentralised if ordinary operators can still run the software. Second, whether developers actually build the on-chain model use cases, starting with options pricing. Third, whether the storage cut lands on the stated timetable, since that is the nearest dated promise in the package.

 

A blockchain planning to double its capacity and host AI models natively is making a clear bet about where demand goes next. Join the discussion at https://zimbuck.com .

 

Frequently asked questions

 

What is the headline target of Solana's 2027 roadmap? Doubling network compute capacity from about 250 million to 500 million compute units per second, alongside an 80% cut in the compute used by token-account transactions.

 

What is Alpenglow, and why does it matter here? Alpenglow is the consensus upgrade activated in October that cut finality from about 12.8 seconds to roughly 150 milliseconds. The 2027 market-structure work builds on it.

 

What does vector processing add? It lets the blockchain's virtual machine handle AI-style maths natively, so models — starting with things like options pricing and margin models — could run fully on-chain.

 

When do the changes go live? The roadmap gives targets for 2027 without a confirmed date for each item. One nearer change, the storage rent cut from 5,080 to 696 lamports per byte, is expected within weeks.

 

What are Rotor and Constellation? Rotor is a planned replacement for Turbine, the block-propagation system; Constellation sets clearer transaction ordering rules.

 

This article is market news and general information only, not financial advice.