The Token Trap

How DappRadar lost its runway overnight.

Author

Nicolás Cerdeira
November 20, 2025

Hey - It’s Nico.

Welcome to another Failory edition. This issue takes 5 minutes to read.

If you only have one, here are the 5 most important things:

Today’s sponsor, Silly Money, helps founders cut their 2025 tax bill with straightforward strategies to keep more of what they earn. Also, shoutout to Cloudways, grab their Black Friday deal for effortless site migrations and big savings.

Stop overpaying on taxes in 2025 AD

Let’s be real: Your biggest expense every year isn’t your tools or your team.

It’s your taxes, and most founders pay wayyy more than they should.

Written by Ankur Nagpal (who sold his last company for $250M), The No BS Guide to Paying Less in Taxes breaks down the exact strategies wealthy business owners actually use:

  • Smarter entities

  • Deduction stacking

  • Retirement tax hacks

  • And moves that 17+ more tactics to cut taxes fast

With December 31st fast approaching, now’s the time to act.

This Week In Startups

🔗 Resources

Host your website on lightning-fast cloud servers and scale without limits. Get 50% off for 3 months plus 50 free migrations with Cloudways. Claim your offer today! *

Collaboration sucks: Be the driver

📰 News

Amazon’s Prime Video is getting AI-generated video recaps for TV shows.

Stack Overflow is remaking itself into an AI data provider.

ChatGPT launches a pilot group chats feature.

💸 Fundraising

AI music startup Suno raises at $2.45B valuation on $200M revenue.

AI customer support agent startup GetVocal raises $26M.

AI cloud operator Lambda raises $1.5B+ to build data centers.

* sponsored

Fail(St)ory

The Crypto Tracker

Seven years of tracking on-chain activity ended with a short post on X: they were shutting it all down. DappRadar became the default tab for a lot of crypto people, and still couldn’t justify a $15k monthly burn. 

The problem was never traffic, it was how the money flowed in and out.

What Was DappRadar:

DappRadar started in 2018 as a simple idea: show what people are actually using on-chain, in one place. It turned raw blockchain noise into rankings and dashboards that didn’t require a PhD to read. Sort of like a Crunchbase for dapps.

The scale they reached was real. By 2025, they were tracking 18,111 dapps across 93 blockchains for roughly 500,000 monthly users. If you cared about on-chain games, NFTs, or dapp usage, you ended up there sooner or later. 

Monetization mostly leaned on their own token, RADAR, and their “subscription” model DappRadar Pro. 

PRO wasn’t a normal subscription though. Instead of paying monthly, PRO users staked RADAR to unlock deeper analytics and alerts.The setup felt clever on paper, but most of the economic weight stayed inside the token rather than turning into cash flow.

They also sold attention to teams looking for oxygen. Banners, sponsored writeups, newsletter spots, social pushes, the usual Web3 marketing menu.

By 2025 they had raised $7.33 million and built a loyal, steady base of users. The problem was the market around them shrinking fast. Crypto funding dropped 70 percent year over year, and their own treasury mostly lived inside a token that kept losing altitude.

The founders said the platform was “financially unsustainable”, and the numbers made the case for them. When the shutdown was announced, RADAR slid another 38 percent, erasing a chunk of the DAO’s remaining value. It turned into one of thspirals where each step speeds up the next.

The Numbers:

  • 📅 Launched: February 2018

  • 👥 Users: ~500,000 monthly

  • 🌐 Coverage: 18,111 dapps across 93 blockchains

  • 💰 Funding raised: $7.33 million total

  • 🔥 Monthly burn: $15.5k

  • 🪙 DAO treasury: $1.6 million total assets; only $46,162 in stablecoins

Reasons for Failure: 

  • The treasury wasn’t real liquidity: Roughly 97 percent of their $1.6 million treasury sat in RADAR, a token already down more than 90 percent. Only 46k was usable stablecoins. When the shutdown news hit and RADAR dropped another 30 percent, the treasury effectively deflated in real time. 

  • PRO didn’t behave like a subscription: Users unlocked PRO by staking RADAR, not by paying in stable value, which meant almost no recurring cash came in. Rewards and lock-up mechanics encouraged holding the token, not funding operations. Great for token optics, terrible for runway. 

  • Their best-paying segments collapsed: DappRadar’s strongest commercial pull lived in gaming and NFTs. Both sectors shrank fast: hundreds of Web3 games dropped off the map and NFT market cap fell almost 45 percent in a month. The budgets they relied on for ads and promotions dried up right when they needed them most.

  • The cost base assumed the bull market would last: Running a multi-chain data index is expensive, and by 2025 their burn was still about $15.5k a month with only three months of stablecoins left. They launched a $249 developer plan, but far too late to matter. Once most of your treasury is an illiquid token, cost-cutting becomes damage control, not strategy.

Why It Matters: 

  • Your token can’t double as your treasury. If most of your runway moves with one chart, you’re not managing finances; you’re riding sentiment.

  • A staking-based membership isn’t revenue. If cash never enters the system, you can’t pay real-world bills no matter how active the product looks.

  • Cycles decide who survives. If your core market shrinks and your cost base doesn’t, the cliff shows up fast and there’s no graceful way down.

Trend

Google’s Busy Week

It was a busy week for Google. Not only did they drop Gemini 3, which now looks like the strongest model on the market, but they also rolled out a new agentic IDE and a serious upgrade to their image generator.

Let’s break down what shipped and why it matters.

Why it Matters

  • Reasoning just leveled up again: Gemini 3’s jump in reasoning isn’t a leaderboard vanity metric. It pushes the ceiling for what you can reliably automate without babysitting. 

  • The IDE war is now about who owns the whole stack: Antigravity signals that Google wants the editor, the agent, and the model under one roof. That kind of vertical control usually compounds fast.

  • Image generation got a big upgrade: Nano Banana Pro behaves more like a proper creative tool: higher resolution, better control, and way fewer retries.

Gemini 3

Google released Gemini 3 on Tuesday, just seven months after Gemini 2.5. The pace alone tells you how aggressively they’re trying to close the gap with competitors. It’s their best model yet, and it landed only days after GPT 5.1, which made the comparison inevitable.

The numbers are solid: a 37.4 on Humanity’s Last Exam (previous high was 31.64) and the top spot on LMArena’s human satisfaction leaderboard. Benchmarks don’t decide everything, but they do shape defaults, and this one will become the new baseline for a lot of teams.

Early chatter is that the model feels more stable and less prone to shortcut answers, though it’s too early to tell where it outperforms in the wild.

Antigravity

Antigravity is Google’s new Gemini-powered IDE. It enters the same arena as Cursor and Windsurf, and the reaction so far is mixed. Some people say it’s exactly what they expected from Google; others say it’s not beating the incumbents yet. Fair enough for week one.

The interesting part isn’t whether Antigravity is better today. It’s how it shifts the category. Google controls the model, the agent layer, and the workspace, and that kind of vertical stack usually wins over time. 

Cursor 2.0 and Warp proved the pattern; Google is now trying to own it end to end. If they ship quickly, this becomes a real threat to everyone building agentic dev tools on top of someone else’s model.

Nano Banana Pro

A couple of days later, Google released Nano Banana Pro, its new image model built on top of Gemini 3 instead of the older Gemini 2.5 Flash. 

The upgrade shows up mostly in control and comprehension, and if you’ve wrestled with earlier image models, you can feel the difference immediately.

It now generates 2K and 4K images; you can dial in lighting, depth, camera angles, and overall scene structure without the model going off-script. It finally handles 16:9 without breaking faces or stretching artifacts across the frame.

And because it inherits Gemini 3’s reasoning, you can actually ask for edits and get revisions instead of a brand new, unrelated image.

Help Me Improve Failory

How useful did you find today’s newsletter?

Your feedback helps me make future issues more relevant and valuable.

Login or Subscribe to participate in polls.

That's all for today’s edition.

Cheers,

Nico