I’ve been watching Solana’s performance story unfold for years, and one question keeps coming back every time a new upgrade lands: how long can a network built for extreme speed keep that edge once serious cryptographic changes arrive? Quantum computing still feels distant to many people, yet the engineering work is already underway. The real challenge is not simply swapping one signature scheme for another. It is doing so while keeping the low-latency, high-throughput character that made Solana stand out in the first place.
The Real Stakes Behind Quantum Preparation
Solana spent years shaving milliseconds off confirmation times and making transactions cheap enough for everyday use. That progress rested on carefully chosen cryptography. Now the same cryptography needs to be re-examined because a sufficiently powerful quantum machine could break the public-key systems currently in use. The threat is not a faster password guess. It is the ability to derive private keys from public information using Shor’s algorithm and then forge valid signatures.
Once that becomes possible, an attacker could authorize transfers without ever controlling the legitimate owner’s seed phrase. That single fact changes the entire security model. Validators rely on signatures to prove agreement. Wallets and applications rely on them to move assets. Both layers need attention, and they cannot be upgraded independently without creating gaps.
In my view, the most interesting part of the current effort is how deliberately the teams are separating different problems. Transaction signatures, validator communication, and migration of existing accounts each require their own approach. Treating them as one giant upgrade would almost certainly damage performance or leave some assets exposed.
Alpenglow Changed the Baseline but Not the Cryptography
Last October the network activated Alpenglow. The upgrade targeted roughly 150 milliseconds for finality under normal conditions, a dramatic improvement over the previous system that often took around 12 to 13 seconds. That is a genuine step forward for traders, payment services, and any application that cares about confirmation speed.
Yet Alpenglow did not replace the signature schemes that secure user accounts or the aggregate signatures that help validators communicate efficiently. It changed how agreement is reached while leaving the underlying cryptographic assumptions largely intact. The voting component is live. The data-distribution component is still planned. Post-quantum adaptations remain research.
This distinction matters more than most headlines admit. A faster consensus system that still relies on vulnerable signatures simply moves the risk somewhere else. Speed and the mathematical hardness of authorization are separate properties. Preserving one while ignoring the other would leave assets open through the remaining path.
For exchanges the difference is practical. Custody software must generate signatures the network accepts. Recovery processes must continue to work. Monitoring systems must understand the new transaction shapes. For application developers a consensus upgrade can stay invisible, while a signature-format change forces integration work across wallets, programs, and signing devices.
Why Signature Size Suddenly Matters
Current Ed25519 signatures occupy 64 bytes. One leading post-quantum candidate, Falcon-512, requires 666 bytes. Another standard, ML-DSA-44, needs 2,420 bytes. Those numbers look abstract until you multiply them by the number of signatures in a busy network.
A single-signer transfer grows by more than 600 bytes under Falcon. A three-signature transaction grows by nearly 1,800 bytes before any other format changes. At one million authorizations the extra data alone exceeds 600 megabytes. Bandwidth is not free. Validators at the edge of their connection capacity feel the difference long before a laboratory machine on a fast local network does.
Transaction size limits make the choice concrete. The existing packet-sized limit sits at 1,232 bytes. A newer format allows 4,096 bytes. An ML-DSA signature already exceeds the smaller limit by itself. Falcon fits more comfortably but still leaves less room for instructions and account references. Larger allowed transactions do not make bandwidth free; they simply shift the cost.
Public keys create a second expense. Falcon-512 keys are listed at 897 bytes versus 32 bytes for Ed25519. Storing a commitment instead of the full key can reduce address overhead, yet verification still needs a secure way to obtain and check the actual key when it is required. Every design decision about where that information lives affects both security and performance.
Falcon as a Practical Candidate
Client teams began testing early Falcon implementations in the spring. Independent work produced similar conclusions: a carefully managed migration could avoid meaningful performance degradation when the time comes. That statement is an expectation, not a completed proof under full production load.
Falcon’s relative compactness is its main attraction among the candidates under serious consideration. A smaller signature leaves more room inside each transaction and reduces the volume of new data that must travel across the network. Verification can be performed with integer-only arithmetic, which is attractive for validators. Producing signatures safely, however, requires careful implementation to avoid side-channel leaks.
Signing happens off-chain, so it does not consume the blockchain’s execution budget. Someone still has to perform it. Wallet providers, institutional signing systems, and hardware devices become responsible for that step. A scheme that verifies quickly on a server may still feel slow or unreliable on a phone, browser extension, or secure element.
Mixed support during the transition creates its own friction. A wallet may understand the new signature while an exchange or connected device does not. Removing the old method too early strands legitimate users. Leaving it available after it becomes forgeable preserves the attack route. Migration planning must therefore identify the exact capabilities required before legacy signing can be restricted.
Quantumglow and Validator Communication
Wallet signatures are only half the story. Alpenglow relies on compact aggregate signatures to represent agreement from many validators at once. That efficiency helps keep communication lean. Suitable post-quantum aggregates that meet the same requirements have not appeared under the constraints the research teams set.
The proposed adaptation, often called Quantumglow, takes a different route. Validators would collect enough evidence and record certificates locally rather than relying on a single compact aggregate. An approval mechanism supports the necessary communication between participants. The design uses hash-based signatures and advance exchange of one-time public-key material, moving some work ahead of the messages that need to be processed quickly.
This approach changes more than the size of a signature field. It alters the communication pattern itself. Evidence is assembled at different times, retained by different parties, and communicated under different conditions. The ordinary path under healthy participation may remain fast while fallback behavior during delays or faults takes longer. Speed during normal operation and the ability to keep making progress under disruption are separate requirements.
A percentage of stake is not the same as a percentage of machines. Fifty small validators and one large validator contribute different voting weight. Any realistic test must disclose the stake distribution, because delaying a large participant affects the fast path differently from delaying several smaller operators whose combined weight is modest.
Independent implementations will be essential. They expose assumptions about networking, hardware, and operator behavior that simulations alone cannot fully reveal. A formal security argument and a practical benchmark answer different questions. Neither replaces the other.
Migrating Existing Accounts Safely
Creating new accounts under a post-quantum scheme is the easier half of the problem. Existing holdings still need a path that remains convincing after the old signature system becomes vulnerable. The proposed approach proves knowledge of the original seed material that generated an Ed25519 account and binds that knowledge to a new post-quantum public key.
The construction exploits a useful distinction. Breaking the conventional signature does not automatically give an attacker the seed knowledge required for the migration proof. That separation is valuable, yet policy still matters. If a vulnerable legacy signature continues to authorize unrestricted spending, an attacker can simply use the old route.
Early prototypes produce proofs around 400 kilobytes. That size is roughly a hundred times larger than a 4,096-byte transaction budget, which immediately shows why migration cannot share the same pathway as ordinary payments. Compression, staged submission, or specialized verification may reduce the practical burden, but the engineering work remains distinct.
Migration is an occasional operation. A large one-time cost can be acceptable when the same cost on every transfer would be impractical. The users least prepared to act, dormant accounts, outdated wallets, and institutional systems with long approval cycles, form an important part of the decision. Selecting new cryptography is often faster than updating every custody arrangement that depends on the old keys.
What Application Vaults Can and Cannot Do
Some teams are exploring application-level vaults that add extra post-quantum checks before assets can leave a program’s control. Assets actually placed inside such a vault can benefit from the additional requirement. The protection is limited to what the program controls and the network assumptions that continue to hold underneath it.
A vault does not rewrite validator consensus. It cannot change every other application that still accepts an elliptic-curve signature as sufficient authorization. Fee-paying accounts remain relevant even when the main balance sits behind stronger checks. The source of the security claim matters as much as the description on a landing page.
More complicated authorization also introduces ordinary software risks. Contract bugs, lost secrets, and approval mistakes remain possible after a stronger signature is added. Wallet interfaces must describe the new approvals accurately enough that users understand what they are signing. Stronger mathematics does not prevent someone from approving a malicious instruction that looks harmless on screen.
Measuring Whether Speed Survives
Claims that performance will be preserved need matching evidence. Fast verification of a single signature does not prove unchanged transaction throughput. A consensus simulation does not prove wallet migration at scale. Useful benchmarks disclose client version, hardware, network conditions, signature scheme, and the exact mix of transfers, trades, and other operations.
Multi-signature workloads deserve separate results because added data scales with the number of authorizations. Systems tested only with simple one-signature transfers can understate the impact on institutional or complex application flows. Operators need CPU, memory, bandwidth, and storage figures alongside finality times. An unchanged confirmation time achieved through much more expensive infrastructure still changes the economics of running a validator.
Ordinary users experience a different set of numbers: signing time, transaction size, fee, failure rate, and recovery compatibility. Publishing those figures next to validator results would show whether performance has truly been preserved across the whole service. Research can establish a plausible path before every implementation choice is settled. Deployment requires the additional decisions that tell operators and users exactly what changes, when it activates, and how existing accounts continue to function.
Looking ahead, several concrete signals will matter more than any single announcement. Activation proposals that identify which post-quantum changes are proposed, approved, implemented in clients, and activated on the live network will provide clarity. Reproducible benchmarks covering finality, bandwidth, signing, verification, and transaction failures under identical hardware and workload assumptions will let outsiders judge the claims. The verification cost, submission method, and recovery rules for binding existing accounts to new authorization will determine how smoothly migration can occur. Wallet compatibility across phones, hardware devices, exchanges, and institutional systems must be visible before any restrictions on legacy signatures take effect. Finally, the infrastructure cost required to preserve fast finality will reveal whether independent operators can still participate competitively.
Solana’s developers have given themselves a preparation path rather than a completed transition. Alpenglow supplies a new performance baseline. Falcon testing, account-migration prototypes, and the Quantumglow research explore different pieces of the remaining puzzle. None of those pieces by itself proves the network has become quantum-safe while keeping its speed advantage. Together they show serious work is underway and that the engineering choices are being made with performance firmly in mind.
The hardest part may not be the cryptography itself. It may be coordinating the long tail of wallets, custodians, applications, and dormant accounts so that the transition does not strand legitimate users or leave attack routes open longer than necessary. That coordination problem is familiar across the industry, yet Solana’s emphasis on speed makes the trade-offs sharper than on slower networks.
I expect the next two years to bring clearer specifications, more independent benchmarks, and at least one serious attempt at limited activation of new signature support. Whether the network can absorb those changes without visible slowdowns will depend on the quality of the implementations and the care taken during the transition window. The work is still early, but the direction is clear: Solana intends to meet the quantum challenge without giving up the performance characteristics that define it. The open question is how cleanly the pieces will fit together once they move from research into production.
For anyone building on the network or holding significant value there, the practical takeaway is straightforward. Watch the activation proposals and the benchmark conditions more closely than the marketing language. Speed claims only hold weight when the full workload, including multi-signature flows and realistic network conditions, is measured and disclosed. Quantum resistance is coming. The networks that preserve both security and performance through the transition will be the ones that treated the engineering constraints honestly from the start.
The conversation around quantum threats often swings between panic and dismissal. Solana’s current approach sits in a more useful middle ground. It acknowledges the long lead time required for cryptographic migration while refusing to treat performance as an afterthought. That balance is harder to maintain than either extreme, yet it is exactly the balance a high-throughput network needs if it wants to remain relevant once quantum-capable hardware becomes a realistic concern rather than a distant research topic.
Whether the final design keeps the 150-millisecond finality target under realistic post-quantum loads remains to be proven in production. The research path is open. The client work has begun. The account-migration problem has a proposed solution. What happens next will depend on how carefully those pieces are integrated and how thoroughly the resulting system is tested under adversarial conditions. For a network that built its reputation on speed, that integration work may turn out to be the most important upgrade of the next cycle.