In 2025, Solana validators and MEV searchers collectively extracted over $400 million in value through priority fee auctions and bundle tips on the Jito block engine. That number isn’t a scandal—it’s a feature of how the network allocates block space. But it reveals something most developers don’t fully appreciate: every block on Solana is a competitive auction, and the participants with better infrastructure capture a disproportionate share of the available value.
Understanding how that auction works—who the players are, what they’re competing for, and what determines the winners—is useful not just for MEV searchers but for anyone building applications where timing matters. If you want to go deeper on who provides the infrastructure that powers this ecosystem, we covered the current landscape in detail here: https://rpcfast.com/blog/solana-infrastructure-providers.
What MEV actually means on Solana
Maximal Extractable Value is the profit available by controlling the order of transactions within a block. On Ethereum, MEV became notorious through frontrunning—bots inserting transactions before and after a large trade to capture the price movement. On Solana, the mechanics are different but the competitive dynamic is similar.
Solana’s 400ms slots mean the window for any given MEV opportunity is extremely tight. A large swap on Raydium that moves the price creates an arbitrage opportunity that exists for one to two slots—400 to 800 milliseconds—before other trades correct it. The searchers who capture that opportunity are the ones who saw the incoming transaction earliest, computed the response fastest, and submitted their bundle before anyone else.
The scale of this activity explains the volume numbers. By late 2025, between 40% and 70% of Solana’s DEX volume on peak days came from automated arbitrage. This isn’t organic trading—it’s an algorithmic ecosystem running continuous competitive auctions inside every block.
How the block auction works
Jito’s block engine sits between searchers and validators. Here’s how the flow works:
- A searcher’s agent detects an incoming transaction via ShredStream or Yellowstone gRPC before it confirms
- The agent computes whether an arbitrage or other MEV opportunity exists and estimates the profit
- The agent constructs a Jito bundle—one or more transactions that must execute in order, atomically
- The bundle includes a tip that compensates the validator for guaranteed inclusion at a specific position
- The Jito block engine receives bundles from all competing searchers and runs an auction: highest tip relative to compute consumption wins the target block position
- The winning bundle executes at the specified position; losing bundles are either delayed to a later slot or dropped
The tip economics are stark. Searchers typically pay 50–60% of expected profit as the Jito tip. A searcher who extracts $1,000 from an arbitrage opportunity pays $500–600 to the validator. At scale across millions of transactions, those tips compound into the $400 million annual figure.
The five things that determine who wins
The MEV competition on Solana isn’t random. Consistent winners share specific infrastructure and operational advantages:
- ShredStream access: seeing incoming transactions at the shred level—before full blocks assemble—provides 50–100ms of additional signal time versus Yellowstone gRPC alone. In a one-slot window, that head start is often the entire margin
- Dedicated bare-metal nodes: shared cloud infrastructure introduces latency variance from other tenants. Dedicated nodes in colocated facilities produce consistent sub-4ms round trips to the leader rather than the 50–200ms range typical of shared endpoints
- SWQoS-enabled submission: bundle tips win the Jito auction, but the bundle still needs to reach the block engine before the slot closes. Staked validator paths protect against intake congestion that can drop transactions before they enter the auction
- Parallel endpoint submission: sending bundles to multiple Jito block engine regions simultaneously—US East, EU, Tokyo—hedges geographic variance in leader proximity. Different validators are closer to different endpoints on any given slot rotation
- Tip calibration loops: static tips lose over time as competition evolves. Production searchers run continuous feedback loops tracking bundle acceptance rates and adjusting tips dynamically based on observed competition levels
The validator side: why this benefits the network
MEV extraction sounds extractive by definition, but the economics on Solana create genuine network benefits alongside the competitive dynamics.
Validators earn 50% of priority fees plus Jito tips. As MEV competition intensified through 2025, validator revenue from these sources grew significantly—providing stronger economic incentives for validators to maintain high-quality infrastructure, reduce orphaned slots, and stay current on client updates. The Foundation’s network health reports through 2025 showed improving validator economics correlated with increasing MEV activity.
The arbitrage activity itself provides a service: price efficiency. When a large swap on one DEX creates a mispriced pool, arbitrage bots correct that price within one to two slots. Users trading against corrected prices get fairer execution than they would in a world without the arbitrage. The extractors and the broader market participants have partially aligned interests, even if the dynamic looks purely competitive from the outside.
What changed in 2025 that made this ecosystem explode
The MEV ecosystem on Solana didn’t reach its current scale overnight. Several developments converged in 2025 to create the conditions for it:
The SWQoS upgrade made transaction submission more reliable for participants with staked validator relationships, reducing the randomness that previously made MEV strategies difficult to run consistently. Local fee markets meant that a congested meme coin launch didn’t flood the entire network—contained fee pressure made it easier to calibrate tips for specific opportunity types. Prop-AMMs proliferated: proprietary automated market makers with dynamic pricing models that create more short-lived pricing inefficiencies than static constant-product AMMs. Jito’s block engine matured, adding regional deployments and improving bundle latency. And the broader AI agent ecosystem gave more teams the tooling to build sophisticated searchers without starting from scratch.
The result is a Solana MEV market that in 2026 resembles a mature HFT environment more than the early experimental arbitrage bots of 2022.
The barrier to entry is infrastructure, not strategy
The arbitrage strategies themselves are well-documented. Cyclic routes, triangular arbitrage, liquidation capture—the logic is public, the Jito documentation is thorough, and the open-source tooling covers most of the implementation work. What separates teams that capture consistent value from those that don’t is almost entirely infrastructure.
A searcher running on a public RPC endpoint with standard WebSocket subscriptions isn’t competing in the same auction as a searcher with ShredStream, a dedicated colocated node, and parallel Jito endpoint submission. They’re submitting into the same block, but the competitive conditions they face are structurally different. The first searcher sees opportunities 100–200ms later, submits with higher latency, and competes from a worse position in the tip auction. On a per-opportunity basis, the infrastructure gap compounds into a large performance gap over time.
This is why the infrastructure layer—not the algorithm—is the primary competitive variable in Solana MEV in 2026.
