Three Launchpad Eras in Five Years: Inside AX1's Console 2.0 Bet on Base

AX1, a startup cluster on Base, has shipped Console 2.0 and put a label on it: agentic activities launchpad. That's the third version of the same idea in five years. The sequence matters more than the branding.

Launchpads keep getting rebuilt because the last one always breaks. Not at launch. Later.

Three Eras, Five Years

Token launchpads came first. Polkastarter opened in late 2020. By 2021 a retail trader could park stablecoins in a pool and walk away with an allocation to a token before it hit any exchange. The mechanic worked. Then it stopped working, because the tokens that came out of it had nothing behind them but a roadmap and a Discord.

Pump.fun in January 2024 finished the job. A few clicks, a few dollars, and the entire pre-list ritual collapsed into a bonding curve. Supply creation went from a privilege to a button press.

Then came agents. Virtuals Protocol went live on Base in early 2024 with a cleaner story. Spin up an AI agent, give it a token, let the market price how useful it might become. The numbers got loud fast. Virtuals' own market cap cleared $4 billion in January 2025. AIXBT, one of the earliest agents launched there, ran past $500 million at its peak. Copycats appeared on Solana, on BNB Chain, on any chain with a grants committee and a Twitter account.

And then the floor gave out. Most of those agent tokens now trade 80 to 95 percent below their January 2025 highs. Some are down 99 percent.

The launchpad wasn't the failure. The supply was. Thousands of agents launched. A handful had a workflow anyone used twice. Numbers in context: at the peak, roughly 15,000 agent tokens existed across chains. Maybe 40 had more than a thousand weekly active users. That ratio is the whole story.

AX1 is aiming straight at that gap. Console 2.0 reframes the launchpad from a token event into a task event. The claim is that an agent launches alongside something it actually does, and that activity gets tracked rather than assumed. Whether the scoring holds up under real usage is the open question. But the framing is correct. Launchpads never failed on fees. They failed on retention.

What Actually Changes

Start with who feels this first. Virtuals. The agent launchpad category belongs to them by default, and every new entrant that reframes the model chips at that default. AX1 doesn't need to beat Virtuals on volume to matter. It needs to make 'agent token' feel like an incomplete unit of account.

Then Base itself. Base has quietly become the chain where agent infrastructure gets built, not just where agent tokens get flipped. That's a meaningful shift for a network that spent 2024 as the cheap place to do retail things. If activities become the unit, Base gets a second act that doesn't depend on memecoin volume.

Who loses? The pure token launchpad. The ones with no activity layer, no scoring, no way to distinguish a working agent from a JPEG with a system prompt. They get commoditized the same way allocation pools did in 2022.

And the LP. This is the part nobody wants to say out loud. If the product is activities, the token becomes a claim on future activity, and the market has no reliable way to price that yet. Visualize this: plot agent token market cap against weekly active agent users over the last eighteen months. The two lines decouple hard starting around February 2025. Market cap kept climbing while usage flattened. That divergence is what the entire third generation of launchpads is trying to fix.

The chart tells the story. It usually does.

Here's my take, and it's not a comfortable one. Most 'agentic activity' in 2026 will be bots farming points for airdrops. That's not cynicism. That's just what happens when you attach a score to a task and a reward to the score. Every activity-based system in crypto has been gamed within weeks of launch. Ax1 will get gamed too. The question is whether the team can iterate faster than the farmers, and whether the scoring survives contact with a few thousand scripts.

Second take. The agent token trade wasn't a technology bubble. It was a supply bubble. The models were fine. The distribution was broken. Anyone building a launchpad that solves distribution deserves attention, even if the current version is half-right.

The Clock Is the Only Honest Metric

What to watch, concretely. Console 2.0 needs to show activity numbers by end of Q1 2026 that survive a skeptical audit. Not total tasks. Unique wallets completing tasks more than once. That number, if it clears five figures monthly, changes the conversation. If it doesn't, AX1 joins a long list of launchpads that shipped a better mousetrap and discovered the mice had already left.

Watch Virtuals too. If they bolt an activity layer onto their existing rails within two quarters, the third generation collapses into the second and AX1 becomes a feature, not a category. That's the real risk. Not competition. Absorption.

One chart, one takeaway: launchpads don't die from bad mechanics. They die from empty follow-through. AX1 has the right diagnosis for a disease that's been running since 2020. Now it has to prove the cure isn't just a nicer dashboard.

Five years, three eras, and the same unanswered question. Does anyone actually use the thing after they mint it?