Solana MEV Enters a New Phase: How Firedancer and AlpenGlow Change the Game
Two major upgrades, Firedancer and AlpenGlow , are reshaping how billions of dollars move through Solana’s network.
Leon Okwatch10 min read·Just now--
Every time you swap tokens on a Solana DEX, a shadow economy activates behind the scenes.
Bots race to re-order transactions in milliseconds. Validators earn extra income beyond ordinary fees. And a small army of specialized traders , called searchers, extract millions of dollars everyday from the gaps in how Solana processes transactions.
This shadow economy has a name: MEV, or Maximal Extractable Value.
In 2026, two of the biggest infrastructure upgrades in Solana’s history, Firedancer and AlpenGlow, will change how MEV works at a fundamental level. A third system, known as BAM, goes even further and attacks the problem at the infrastructure layer.
Whether you’re a DeFi power user, a casual trader, or just someone holding SOL and watching the ecosystem grow, this is the context you need to understand what’s actually happening under the hood.
What MEV Actually Is
Think of it this way. You submit a swap transaction. It sits briefly in a queue before a validator picks it up and adds it to a block. During that window, sophisticated bots can see your pending trade and react.
They place orders before yours, after yours, or both, capturing profit from the price movement your trade is about to cause. You still get your swap. You just get it at a slightly worse price, and someone else pocketed the difference.
That is MEV in its most common form. On Ethereum, researchers have documented and debated it for years. On Solana, the mechanics differ slightly because there’s no global public mempool. But the core dynamic is identical: whoever controls transaction ordering controls the money flow.
Before the current infrastructure existed, Solana’s MEV environment was pure chaos. Bots competed by spamming the network with thousands of copies of the same transaction, hoping one would land first.
At peak congestion, over 60% of Solana’s block compute units went to arbitrage attempts, and more than 98% of those attempts failed. The network was burning through most of its capacity on a machine-gun lottery.
Validators earned nothing extra for the trouble, users faced degraded performance, and the value being extracted leaked out of the ecosystem rather than flowing back to infrastructure providers.
How Jito Fixed the Chaos
Jito Labs, founded in 2021, reframed the MEV problem as an auction problem.
Instead of letting bots spam the network, Jito built an off-chain Block Engine, a matching system that accepts transaction bundles from searchers, simulates them to check profitability, and forwards the most valuable ones to validators running the Jito-Solana client.
Searchers attach SOL tips to their bundles as bids. Validators take the highest-paying bundles. Everyone stops wasting compute on failed attempts.
The economic result was significant. By the end of 2025, MEV revenue on Solana crossed $720 million for the full year, more than all priority fees combined, making it the single largest component of Solana’s Real Economic Value (REV) for the first time.
Since its launch, Jito’s infrastructure had paid out roughly $674 million to stakers and validators. Nearly 94% of staked SOL now sits on validators running Jito software, because validators on Jito earn meaningfully more than those that don’t, and stake follows yield.
But Jito’s dominance also created a structural risk. When 94% of validators run the same underlying codebase, a single critical bug threatens the whole network.
More immediately, Jito’s public mempool was enabling sandwich attacks at scale. In one 30-day window, a single program executed 1.55 million sandwich transactions and extracted roughly $13 million from retail users.
In March 2024, Jito shut down its public mempool. It sacrificed short-term revenue to protect the users the network depends on. Sandwich attack frequency fell sharply.
But the move created a secondary problem: private, permissioned mempools started filling the vacuum, less transparent, and accessible only to those with the right connections.
Firedancer: A Completely Different Kind of Validator
Solana’s current validator software, Agave, can theoretically process around 65,000 transactions per second.
Firedancer, a new validator client built from scratch by Jump Crypto, handled over one million TPS on standard hardware during demonstration testing. That’s not an incremental improvement. It’s a different order of magnitude.
Jump’s engineers spent three years building Firedancer in C and C++ — languages optimized for raw hardware performance.
Where Agave runs as one large monolithic program, Firedancer splits every function into isolated “tiles.” One tile handles networking. Another handles signature verification. Another handles transaction processing.
They run in parallel and communicate through shared memory. If one tile crashes, the others keep running. The failure stays contained.
The rollout happened in deliberate stages. Frankendancer, a hybrid combining Firedancer’s high-performance networking with the existing Agave runtime, went live on mainnet in September 2024.
Validators running Frankendancer were actually leaving MEV revenue on the table, because Jito’s bundle auction wasn’t yet integrated. Despite that, Frankendancer’s stake share jumped from 8% to over 20% in just four months. The validator community voted with their hardware for network resilience, not just yield.
Full Firedancer reached Solana mainnet in December 2025, after more than 100 days of continuous testnet operation across 50,000+ blocks without major incidents. Within weeks, it was running on more than 20% of active validators.
For MEV specifically, Firedancer’s higher throughput changes the economics of blockspace in a direct way.
More block space means less competition for inclusion. When a single block can hold ten times as many transactions, the scramble to get in becomes less intense, which naturally deflates the extreme fees and tips that searchers need to pay.
The total dollar value of MEV may keep rising (more activity), but the cost per transaction should fall. More room in each block means less competition for inclusion. When a block can hold far more transactions, searchers and users don’t need to bid as aggressively to guarantee timely processing.
The total dollar volume of MEV activity may keep growing as Solana’s ecosystem expands, but the cost per transaction should come down.
Meanwhile, Jito is actively integrating its BAM system with both Agave and Firedancer clients, which will close the MEV revenue gap and accelerate Firedancer adoption further.
AlpenGlow: Finality in the Time It Takes to Blink
Solana is already considered fast. A swap typically confirms in roughly 400 to 800 milliseconds — far quicker than Ethereum’s minutes. But that window is still long enough for a bot to react, for a price to move, and for a predatory trade to land between yours and the market.
AlpenGlow targets finality in under 150 milliseconds. That’s roughly the duration of a single blink. At that speed, Solana stops being just a fast blockchain and starts approaching the responsiveness of the web2 apps on your phone.
Solana’s developer community announced AlpenGlow at the Accelerate conference in May 2025. Validators passed a governance vote in September 2025 with a 98.27% yes vote, one of the strongest community mandates in Solana’s history.
The protocol is expected to go live later in 2026.
The implications for MEV are structural. A significant portion of MEV strategies depend on a time gap between transaction submission and finalization.
Bots use that gap to observe pending state, construct profitable bundles, and insert them ahead of or alongside the target transaction. When that window collapses to 150 milliseconds, many of those strategies simply can’t execute fast enough to be profitable.
AlpenGlow doesn’t just improve user experience, it removes the timing advantage that some forms of predatory MEV depend on, but bots with the the best infrastructure may still be profitable.
At the same time, faster finality creates new opportunities. Market makers can quote tighter spreads because their inventory risk window is shorter. Arbitrageurs can cycle capital faster, compressing price discrepancies across DEXs more efficiently.
Liquidators can call undercollateralized positions faster, reducing bad debt risk in lending protocols. The net effect is a tighter, more efficient market that benefits participants who add genuine value rather than those who exploit timing gaps.
BAM: The System That Actually Kills Sandwich Attacks
Of all the changes in Solana’s infrastructure stack, the Block Assembly Marketplace, BAM, has the most direct impact on ordinary users. Jito launched BAM on Solana mainnet in September 2025.
BAM introduces an encrypted mempool.
Today, even with Jito’s public mempool shut down, transaction data still travels through systems where well-positioned observers can see it before it’s finalized.
BAM closes that gap using Trusted Execution Environments (TEEs), which are specialized secure hardware enclaves that process transaction data in a sealed environment. The operators can’t see it. The validators can’t see it. Bots definitely can’t see it. Your transaction stays encrypted until the moment it executes on-chain.
In practice, here’s how it works:
Sandwich attacks don’t just become harder under BAM. They become cryptographically impossible for participating validators.
BAM consists of three interconnected layers:
BAM Nodes are the hardware backbone, essentially servers running inside TEEs that receive transactions directly from RPCs and validators, sequence them privately, and generate cryptographic proofs of their ordering.
BAM Validators run the updated Jito-Solana client and receive pre-ordered, cryptographically attested transaction sets from BAM Nodes for execution.
And Plugins, the most architecturally interesting piece, are programmable interfaces that let developers build custom transaction ordering logic directly into the system.
This plugin system, which Jito calls Application-Controlled Execution (ACE), opens design space that didn’t previously exist on Solana. A central limit order book exchange can build a plugin that ensures its orders fill fairly and efficiently before settling on-chain.
A lending protocol can manage liquidations in a defined sequence. A payment processor can guarantee ordering guarantees for its users. These are capabilities that previously required a layer-2 solution. With BAM, they run natively on Solana mainnet.
The economics work differently too. Plugins generate fees when used. Those fees distribute across BAM Node operators, validators, stakers, and the Jito DAO. Under governance proposal JIP-24, passed in August 2025, 100% of fees from both the Block Engine and BAM flow to the Jito DAO Treasury.
By Q2 2025, before BAM’s full mainnet launch, the existing Block Engine was already generating roughly 22,391 SOL, about $4 million, for the DAO through its Tip Router mechanism.
As BAM matures, that number grows alongside plugin adoption.
What the Risks Look Like
These upgrades are genuinely impressive. But they come with open questions that deserve honest acknowledgment.
- Centralization remains the biggest concern. Jito’s stake dominance is economically rational for individual validators, but it creates systemic risk if Jito’s infrastructure fails or makes decisions that conflict with the broader network’s interests. BAM adds new trust dependencies on TEE hardware manufacturers. Jito has committed to expanding BAM Node operations, open-sourcing the codebase, and transitioning governance to the DAO, but the timeline and execution of that decentralization matters enormously.
- TEEs are not perfectly secure. Side-channel attacks and supply chain vulnerabilities have been demonstrated against major TEE implementations. If a TEE vulnerability were exploited in BAM’s context, it could re-enable the frontrunning that BAM was built to prevent. The cryptographic attestations that make BAM trustworthy are only as strong as the hardware they run on.
- AlpenGlow has never run at production scale. Replacing a core consensus mechanism on a network with $700 million in annual MEV activity is among the most ambitious changes any major blockchain has ever attempted. Unexpected behavior under production load could delay mainnet activation, or worse, require a rollback.
- Private mempools are filling the vacuum. Jito’s 2024 public mempool shutdown created space that private, permissioned ordering systems have begun to occupy. Unlike Jito’s public system, these operate without transparency and primarily benefit participants with exclusive access. BAM is the structural answer to this problem, but it only works as a fairness mechanism if it achieves widespread validator adoption before private alternatives become entrenched.
The Bigger Picture
Firedancer, AlpenGlow, and BAM don’t exist as isolated upgrades. They interact.
Higher throughput from Firedancer makes AlpenGlow’s faster finality more economically meaningful, and blocks stay full and valuable even at higher speeds.
Faster finality from AlpenGlow compresses the timing windows that make MEV auctions gameable. And BAM’s encrypted sequencing layer enables the kind of application sophistication that drives the transaction volume that makes MEV revenue worth competing for in the first place.
The era of brute-force MEV — spamming the network, sandwich attacking retail users, front-running trades through observable pending state — is structurally narrowing for Solana’s core infrastructure.
What replaces it is a system of verifiable, auction-based blockspace markets that distribute value more broadly. Validators earn from structured tips. Stakers earn from DAO fee flows. Developers earn from plugin usage. And users get transactions that are faster, cheaper, and provably harder to exploit.
MEV doesn’t disappear in this world. The economics of blockspace guarantee there will always be value in transaction ordering. But the nature of that value shifts , from extraction built on information asymmetry and timing exploitation, to structured competition in fair, transparent auctions.
The question Solana’s community faces now is whether its governance, decentralization safeguards, and competitive dynamics can keep pace with the economic forces concentrating around the infrastructure layer.
The technical vision is coherent and ambitious. The execution, as always, is what determines whether it holds.
