Why Solana Token Launches Fail in the First Block

Most tokens that fail on Solana do not fail because of the idea, the branding or the community. They fail in the first four hundred milliseconds, before anyone outside the launch has seen them exist.

A block on Solana lasts roughly that long. When a token is created, the moment it becomes tradable is written to a public ledger, and automated buyers are watching the chain for exactly that event. They see the mint before any human does, and they buy at the lowest price the token will ever have. By the time the creator posts about the launch anywhere, a meaningful share of supply already sits in wallets that have no interest in the project beyond selling it back at a markup.

This is the problem that bundling solves, and it is worth understanding what the term actually means, because it gets used loosely and often as an accusation.

What actually happens in block zero

Solana processes transactions in discrete blocks. Jito, a block engine used by most validators on the network, accepts groups of transactions called bundles and either includes all of them in the same block or none of them at all. That atomicity is the entire point of the mechanism.

A launcher can place the token creation instruction and a set of buy transactions into a single bundle. If the bundle lands, the buys settle in the same block as the creation. There is no window for anyone outside to react, because from the perspective of the network the token and its first buyers appeared simultaneously. Nothing preceded anything else.

The practical limit is five transactions per bundle. Each transaction can carry several buy instructions, which is how a launch spreads across a dozen or more wallets while still fitting inside one block. Anything beyond that ceiling lands in a later block, after the token is publicly visible, and is no longer part of the bundle in any meaningful sense.

Bundles also carry a tip to the validator. Higher tips improve the chance of inclusion but guarantee nothing, because bundles compete against each other for limited space. A tip that worked reliably last week may be insufficient during a period of heavy network activity.

Why this is not automatically manipulation

Bundling has a reputation problem, and some of it is deserved. A creator who takes forty percent of supply across fifteen wallets and sells into the first wave of organic buyers is running a scheme, whatever they choose to call it. Those launches are common enough that the word itself now carries suspicion.

But the same mechanism is used defensively, and the defensive case is not rare. Without it, sniper bots take the opening supply and the creator holds nothing in their own project. Teams that intend to hold, build and stay face exactly the same automated buyers as everyone else. Doing nothing is not a neutral choice in that environment; it simply hands the opening to whoever has the fastest infrastructure.

What separates the two cases is not the technique. It is how much supply ends up concentrated and what those wallets do in the following hours and days. A launch where block-zero wallets are still holding weeks later looks nothing like one where every wallet is empty within the hour, even though the first block of both is structurally identical.

Reading a launch after the fact

Anyone can verify this without special access. The creation block is public, every transaction inside it is public, and the current balance of every wallet involved is public.

Three numbers matter. How many distinct wallets bought before the token was visible. What share of total supply those wallets took. And how much of that supply they still hold today.

The third number is the one most people skip, and it is usually the most informative. Supply that has already been sold went to someone, at a higher price than the block-zero buyers paid. If the wallets that opened a token are empty, the distribution has already happened and the people holding now are on the other side of it.

Tools differ in how they measure this, which is why two checkers can report very different figures for the same token. Some read only the single transaction that created the token and miss buys that settled in separate transactions within the same block. Others count the bonding curve or liquidity pool as a buyer, which holds nearly the entire supply and produces figures above one hundred percent. Reading the entire block and excluding program-owned accounts gives a number that matches what actually occurred on chain.

For anyone planning a launch

The mechanics are not the difficult part. Wallet generation, funding, deriving the mint address in advance, building transactions and submitting them as a bundle are all automated by tools like SolBundler, which handles up to twenty wallets in a single Jito bundle and manages the post-launch work that most standalone scripts ignore entirely.

Collecting dust from twenty wallets, tracking what each one holds and selling in a controlled sequence are all part of the same job. Plenty of scripts get a bundle into block zero and stop there, leaving the operator to manage twenty keypairs by hand afterwards. That is where the hours actually go.

The decisions stay with the launcher regardless of tooling. How much supply to take, how to distribute it across wallets so it does not read as a single entity, what tip is worth paying, whether the concentration is defensible if someone checks. No software makes those choices, and they determine the outcome far more than the execution layer does.

What no honest tool will promise is guaranteed inclusion in block zero. Bundles compete for space, validators can and do reject them, and network conditions change hour to hour. Anyone claiming certainty on that point is describing marketing rather than how the network actually behaves.