Use case #2: an OTC token sale, run by a DAO.
Price, whitelist, per-wallet limits, cliff and vesting β all set in one proposal. Nothing leaves the treasury until members vote it through.
This is what DeXe was built for. #DeXe#DAO
Build log, part 1.
Three months on the foundations of the DeXe platform. Numbers below β measured, not estimated.
Platform stats used to take 7β12s to load. We rewrote the heaviest query behind them: 20.2s β 0.96s on the database. First load takes a few seconds if stats haven't been requested recently; after that, ~0.1s.
The app got lighter: main page network requests 140 β 60, worst scroll stall in the DAO feed 247ms β 0. Measured in the browser by an automated check, not by eye.
The bugs you actually hit: WalletConnect login dead-ends β fixed. Spam NFTs β filtered out of wallet views (7 of 53 on the wallets we checked, zero false positives). Network mismatch β a clear prompt instead of a blank page.
The data layer: indexing unblocked for 128 BSC pools, data reads now fail over automatically instead of breaking the page.
Underneath it all runs a monitoring layer: 34 checks per pass, one summary, read-only by design. The checks run in code. Our in-house AI agent only writes up what the checks find. Monitoring rule: a data source going quiet is never proof of health.
That was part 1. The build log continues.
#DeXe
@ylmazqeps@DexeNetwork The actual $DEXE in circulation is only 48 million, balance all are either staked or locked.
Definitely the price will be manipulated and it will go 10-15X soon
π $DEXE : Smart Money Is Accumulating While Fear Takes Over
The market is focused on the recent price drop, but the on-chain data tells a different story.
Bullish Signals:
DEXE holders just reached a new all-time high of 54,680, showing that more investors are accumulating $TAO
DePIN narratives are opinions. Revenue is the scoreboard.
$IO ranked #1 in 24H revenue at $30.8K - more than $FIL, $AKT and $RENDER combined.
The market may still be debating decentralized compute.
@ionet is already monetizing it. π΄
A 2.8T parameter model launches. Demand spikes. GPU capacity hits its ceiling in 48 hours.
Moonshot had to pause new subscriptions on Kimi K3, not because interest faded, but because it exploded.
Even a $20B company with a leading model is capacity-constrained.
Model quality isn't the bottleneck. Compute is.
Centralized infrastructure can't scale on demand. It scales on procurement cycles, data center buildouts, and capital timelines.
Distributed GPU networks are the answer.
The next Kimi K3 moment is coming. We're building the infrastructure to meet it.
https://t.co/bzDcmhrVIX
$250K β $1M in monthly network earnings. March to May 2026.
The Incentive Dynamic Engine (IDE) ties every dollar of that growth to a burn mechanism.
$12.11M projected ARR. #1 in DePIN revenue.
Full breakdown below in a new report from A1 Research.
You shouldn't need a waitlist to train a model.
You shouldn't be capped by one company's GPU inventory.
You shouldn't pay hyperscaler margins for compute that's sitting idle somewhere else.
Compute is a global resource.
https://t.co/WLXlHkv6f1 treats it like one.