The argument in one line.
Flat-rate AI subscriptions were always subsidized Ponzi math, and the shift to usage-based token billing will create a permanent intelligence class divide between those who own their compute and those who rent it.
Read if. Skip if.
READ IF YOU ARE…
- You pay $20-$200/month for a frontier AI plan and your work depends on that access continuing at the same price.
- You are a solo builder or freelancer whose projects rely on cheap vibe-coding or AI automation.
- You work in trades or small business and have been watching AI from the sidelines without acting.
- You have heard warnings about AI pricing shifts and dismissed them as hype or self-promotion.
SKIP IF…
- You already run local models on high-RAM hardware and have moved past subscription dependence.
- You want a calm, sourced analysis of AI pricing trends — this is an urgent rant, not a structured report.
The full version, fast.
Richard Echols argues that Claude Fable 5's launch signals the end of the subsidized AI subscription model: after June 22, frontier models move to pay-per-token pricing, making what cost $200/month potentially cost thousands. He frames the response as three moves — own local hardware to run open models, build enough skill depth to know when expensive frontier models are worth it, and stop trusting cloud vendors to keep prices accessible. The argument is backed by Alex Finn's X thread and punctuated with his own RMDW product as a live example of acting before the window closed.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.
00:00 – 00:43
01 · Cold open
Jun Song X post read aloud introduces the permanent underclass framing.
00:43 – 01:40
02 · Direct address
Third video of the day, unpolished, urgent: the era of AI subscriptions is over.
01:40 – 02:09
03 · Hold up the hardware
Host grabs a local device off-camera — the hardware thesis made physical.
02:09 – 03:05
04 · Alex Finn thread
Reads Alex Finn's X thread: usage-based billing incoming, subscriptions dead, local LLMs rising.
03:05 – 04:25
05 · Prior warnings + RMDW intro
Defends repeated hardware warnings; introduces RMDW as proof that early action compounds.
04:25 – 05:47
06 · The vendor playbook
Hook on frontier models, nerf subscription tiers, then force token billing.
05:47 – 06:50
07 · Apple local model demo
Shows Apple Foundation Model via fn+Chat; even on-device AI has token limits.
06:50 – 08:10
08 · Mac Studio eBay listing
Reveals he listed a Mac Studio on eBay but is reconsidering; personal Fable usage story.
08:10 – 09:12
09 · Golden age is dying
The $200/month era of unlimited Claude Code instances and usage resets cannot continue.
09:12 – 10:27
10 · Who can build now
Skill depth is the real moat; Opus version arc as drug-and-withdrawal metaphor.
10:27 – 12:30
11 · Hardware as hedge
Three Mac Studios; June 22 deadline; get hardware immediately.
12:30 – 13:55
12 · Wealth gap expansion
Those spending $10K/month on Fable will eat the economy; government intervention speculation.
13:55 – 15:02
13 · Redirect your spend
Cut Netflix, DoorDash, Uber; invest that money in AI instead.
15:02 – 16:32
14 · RMDW client showcase
Maisons Construction case study: client does not touch website, AI handles leads — the frog-in-pot close.
16:32 – 17:10
15 · Use Fable now
Urgency close: maximize Claude Fable before June 22.
17:10 – 18:57
16 · Final close + product demo
RMDW Construction Assistant demo ($99/month); OpenAI and Google will follow; IPO profitability is the forcing function.
Lines worth screenshotting.
- Flat-rate AI subscriptions were never sustainable: companies were selling thousands of dollars of compute for $200/month, and that math only works while investor subsidies flow.
- Usage-based billing selectively prices out the people with the least financial cushion, who are also the ones most dependent on cheap AI access.
- Apple putting a local model on MacBooks while still throttling tokens per device is the tell: the billing gate is always somewhere, even when the model is free.
- The gap is not just money: it is the compounding skill gap between people who used subsidized access to get good and those who dismissed the warnings.
- A $200/month Claude subscription in ultra code mode can burn through its limit in under an hour; the generous plan was always a usage illusion.
- When a company cannot make money on subscriptions, IPO pressure forces them to find profitability — and the only lever left is usage pricing.
- Running a local model you own is the only form of AI access that cannot be price-hiked or shut down by a vendor decision.
- The trades and small business market is early and underserved: a contractor using AI for lead capture has no local competition yet.
- Vibe-coding an app and getting it to reliably work are different skills — the former is being commoditized while the latter compounds with daily practice.
- OpenAI and Google will follow Anthropic's pricing model because their investors are demanding a path to profitability as IPOs approach.
Subsidized access ends; skill depth and owned compute are what remain.
WHAT TO LEARN
The flat-rate AI subscription was a customer acquisition subsidy, not a pricing floor — and when the subsidy ends, access stratifies by who built real skills and who just rode cheap tokens.
- Flat-rate AI subscriptions were priced at a loss to drive adoption; the move to token billing is not a surprise, it is the second act of the original business model.
- Vendor lock-in through AI workflows is real — the deeper your stack depends on a specific frontier model, the more pricing leverage the vendor holds.
- Running a local model you own is the only form of AI access with no pricing floor — open weights can be downloaded once and used indefinitely without recurring cost.
- The skill gap compounds: people who used cheap access to build deep expertise are years ahead of those who stayed on the sideline waiting for prices to stabilize.
- Apple charging tokens on a local model running on your own device is a preview of where every hardware-software integration is heading.
- For trades and small-business operators, AI automation is still early enough that first-mover advantage is real and defensible, since most competitors have not started.
- Knowing which tier of model a task actually requires is itself a skill — using a frontier model for tasks a free local model handles wastes the cost advantage of the frontier tool.
Terms worth knowing.
- Usage-based billing
- Pricing model where you pay per token consumed rather than a flat monthly fee, making heavy use potentially far more expensive than a subscription.
- Frontier model
- The most capable, most expensive AI model a provider offers — distinct from older, cheaper models kept on subscription tiers.
- Local model
- An AI model downloaded and run entirely on your own hardware, with no API costs, no vendor rate limits, and no subscription required.
- Token
- The unit of measurement for AI compute — roughly 0.75 words. Both input and output are billed in tokens under usage pricing.
- Vibe-coding
- Generating working-ish code by prompting an AI without deep understanding of the output — fast to start, brittle under real conditions.
- RMDW
- Real-world Models, Data and Workflows — the host's AI consulting and product business, building vertical AI apps for trade and small-business clients.
- Prompt doom
- The spiral of increasingly expensive prompting when a cheaper model fails, forcing escalation to pricier frontier models to complete the same task.
Things they pointed at.
05:47toolApple Foundation Model (on-device, fn+Chat)
11:20toolGLM open-source model (runs on Mac Studio)
15:00linkMaisons Construction (RMDW client)
Lines you could clip.
00:37
“Welcome to the permanent underclass.”
01:04
“The era of AI subscriptions — what I've been warning you all about — it's over with. It's through. It's cooked. It's done.”
02:45
“The subsidies were just a Ponzi scheme.”
08:10
“The golden age of paying $200 a month and being able to code 40 Claude Code instances and getting a usage reset every five minutes are about to die.”
10:03
“They drug you up, they make you feel real good and then they're gonna take it away.”
16:14
“You're the frog getting cooked in the slow cooking pot that just boils a little bit at the time because you cannot see what's going on right in front of you.”
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
metaphoranalogystory
The release of Claude Faber five just accelerated the intelligence class divide, upgraded intelligence, expensive APIs, heavier censorship, and subscription limits that drain way faster.
The five hour limit on the $200 plan is gone in less than an hour. I personally experienced this.
Moving forward, the productivity gap between devs who use this and those who can't afford it will be dozens of times wider. This applies to companies too.
Microsoft previously banned Opus due to the cost, but now this gap will force everyone to use much more expensive frontier models. Welcome to the permanent underclass.
Hey, man. Y'all look, I'm not even gonna make my face big for this video because I just had to get back on here and record this through videos in a day.
It's not my usual polished up one, but I just gotta tell you, man, it's over. If you didn't listen, it's over.
And, yeah, it's done. The the era of AI subscriptions, what I've been warning you all about, it's over with.
It's through. It's cooked. It's done.
We're here now. We're here now where it's so expensive that the gap is gonna be huge.
Those who can pay for models like Claude Fable and those who can't.
And it's it's just boy, let me let me go grab something. Okay?
Let me go grab something. I'll be right back.
You see you see, I told you all, hey, go buy one of these.
Get the hardware while you can. You know, if somebody's selling the hardware, get it. Why?
Because the time is gonna come when you will not be able to run AI, when you'll not be able to use it. Okay?
I think Alex Finn says it a lot better than I can. Alright? And I'm gonna read his whole thing and I'm gonna get my take on it.
AI subscriptions are dead. Claude Fable five will be will be on the Anthropic subscription until June 22. It all ends June 22.
After that, you'll need to pay for usage per token. This will be the start of a much larger trend. Frontier models will no longer be included in subs.
You'll pay a fee and it will only get you access to older much cheaper models. If you want access to that, uh, dank AI sour diesel, you're going to need to pay for every token you use.
No more subsidies. And that makes sense. The .
For those who don't know, when you pay $200 a month for an AI sub, you get thousands of dollars of tokens. I've been telling you all this in video after video after video. Prepare yourself, get ready, learn from someone like me.
And it's not just the, buy my product. I keep seeing comments. Oh my god, Richard, your videos are just a big infomercial for RMDW.
Oh, you're and y'all look, man. A lot of the people who are here who are supportive, y'all understand, y'all get it.
Okay? I work with a lot of you all. But the ones who aren't here, I told you these guys and girls, whoever it is telling you that people like me are wrong, telling you that people like Alex Finn are wrong, telling you, oh, don't I had a guy, oh, do not buy the hardware for local model.
Well, guess what, man? June 22, the best cloud model is usage based token billing.
So now everybody who did not pay to go get the hardware, now everybody is cooked because because oh, promotion Richard, stick with me.
Okay? Stick with me. Because now something like my business, RMDW, AI where it's what's your business, the biggest pain you want, solve, and then you just click your business and then boom, you know, well, let me go let me go here.
Okay? You just click whatever domain your business is and then there's an app for it. Well, guess what?
Because you slept, Richard was able to make something like this and you won't be able to.
You won't be able to because you were in the comments, oh, yeah. I'm not gonna buy any hardware.
I'm not gonna do this. I'm not gonna do and the era is over. The era of, oh, well, I can just vibe code and app up for anything I need.
No. You can't. No.
You can't. Because this is the play that they're gonna do.
Okay? I'm I'm letting you know ahead of time what's gonna happen. They're going to put token based billing as their highest thing.
So Claude Fable is just the beginning. And then for everybody paying $2,102,100 dollars a month, they're going to nerf you like crazy.
They're going to nerf Opus. They're going to nerf GPT 5.5.
They're going to nerf it all so that you get in the vortex of prompt doom and then you end up paying them usage based billing too.
That's the game. But if you have one of these, if you invest in one of these, if you have the opportunity to, then it doesn't matter because a local model runs that you have downloaded, that is yours, that is private, and then boom, you're good.
Let me tell you why else things are dead. I just updated my MacBook. Just look.
I type f n, Boom. Mac has now put their own local model on their devices. Okay?
I do f in chat. Come in here.
Hello. How are you today?
Are you able to build, or is that update coming later?
And look, yes, I'm able to assist with building and updating models, update for new features, are typically released periodically, so I recommend checking website, all this.
You see, the the the move is we'll put the local model on their computer and then we'll jack up the price of the computer too. And look, this is this is a local model that Apple has put on my computer, but look at that.
They're even me getting for they're even getting me for the tokens on the local model put on my Mac.
Again, if you have a 128 gigabyte Mac or you go get you a Mac Studio or two, you won't have to worry about this.
I'm telling you, I got I got this up for sale on eBay right now. Okay?
And these prices are way too low, but I gotta get it I got it up for sale right now. Guess what? In six days, it's gone.
But guess what, man? I don't really want anybody to buy it from me. I really don't want anybody to buy it from me, especially now because I see the play.
It's it's kinda like, hey, man. I might need to take this down because I need this. I need this so that I can run a frontier model from my home and not have to do usage based billing because I've shown you on video after video the price of paying usage based billing.
It's crazy. You will find yourself paying a thousand, $2,000 in a matter of days.
Look, man. I was using Fable today. Put it in ultra code mode, had a workflow, my usage was gone.
Why do you think Claude resetted everybody's usage today? Because they saw. Oh, man.
This thing is killing people. It's not a good look. The golden age of paying $200 a month and being able to code 40 Claude code instances and getting a usage reset every five minutes are about to die.
The party couldn't continue forever. You can't just leverage the entire global economy for years and expect nothing to break.
Now it's time to pay up. This means a few things. Time to be responsible when it comes to which models you use.
You don't need Fable five or GPT 5.5 extra high for everything. Build the skill of knowing when to use cheap models. Let me tell you where you do need it.
Let me tell you where you do need Fable and GPT 5.5. If you're trying to build a business, if you're trying to build an app.
I should be thankful because Anthropic just made me a lot of money today. Because when they nerf these other models, you're not gonna be able to just vibe code something up and oh, yeah, it works right.
No. That's how they get you. You think you can make a working app, but you can't.
You can't unless you sit here and you study and you're in it day after day like I am, like Alex Finn is. I'm in it because my day job forces me to be in it, which I love, and I do it outside because it's my whole biz the whole way that I am able to feed my family is with work in AI, both from a automation standpoint and both from a building standpoint.
Automation, my nine to five, and then building my own business. But now this is being taken away from you.
If you saw what they did to Opus four point five, which was fire, to Opus four point six, which was, ah, and then four point seven, which was horrible, and then 4.8, which was, wow, we got something.
And now Fable, oh, we got something. . And I'm telling you, you think $20 a month, $100 a month, $200 a month is gonna get you something.
Think about it, man. You're running your whole business off of Kwan because you didn't listen.
You didn't go out and get the stuff. You didn't hire Richard. You didn't no, man.
I'm gonna do my own thing and I'm gonna and these cloud companies, these same guys who love they they they love to get up and say how they wanna fire all of you. They love to get up and say how they want all of you not to be able to take care of your family.
And these are the guys who you have put your trust in because, oh, Alex Finn is just a YouTuber.
Who is he? Oh, Richard is just a small YouTuber.
Who is he? Oh, these guys don't know what they're talking about. No.
I don't wanna talk. No. I don't want and it's like, alright.
Here it is right here. Look at point number two. He says, local LLM slash hardware will come even more in demand.
I'm currently running GLM on my Mac Studio. It's great. Is it Fable?
No. But it gets the job done for free on simple task. Learn about local LLMs.
Do you know what Apple didn't release yet this week? I ain't seen nothing about a new Mac Studio.
Hey, man. WWDC is gone is going on for the rest of the week.
Maybe they'll come out with something this week. Maybe. Maybe.
Maybe not. I don't know. But I but I know this.
I know I'm thankful that I have mine. I know I'm thankful that I have three Mac Studios now.
Like, I'm real I man.
I might have to take that listing off of eBay, like, for real. For real. Because it's over.
June 22, it's over. Get your hands on some hardware immediately.
Immediately. It's it's y'all don't understand.
A lot of y'all do. A lot of y'all do. But there are some who are watching me who have no idea.
They have they have conditioned everyone to be a 100% dependent on AI.
And they condition people who actually are in AI psychosis, who worship this this stuff and and makes them go do things.
That's how deep this has gone to where they're now all IPO ing because the money's dried up, the returns aren't there.
Let me just look. I'll read it.
Alex says it better than me. This is the beginning of the wealth gap expansion. Those that can spend 10,000 a month on Fable will build incredible products that eat up more and more of the economy.
Those that can't afford Fable five will have an insane disadvantage. The government will need to step in eventually. There will be too much civil unrest.
Hey, I I don't know. We'll see. We'll see because a lot of people a lot of people just know AI by chat gbt.
There's a there's there are billions of people right now who don't have any idea what's going on and they don't think it'll impact their life. They don't think that this means anything at all and it and it means a lot.
It means a lot. Reminds me of something else I do but I'm not gonna get into that. Alright.
You need to seriously reconsider where your money goes every month. If you're complaining about AI prices and in the back of your mind, you know your skill set is becoming quickly irrelevant all while spending money every month on Netflix, Xbox, Paramount, drugs, DoorDash, Uber, other things that bring nothing positive to your life, you are simply doing it wrong.
AI is an investment in yourself. It's an investment in your relevance to the global economy. You need to make sure you make that investment.
This is where I'm hey, man. It's it's my whole YouTube channel.
It's my whole YouTube channel. If you don't learn AI, it's gonna be real hard for you.
I work with guys in the trades. They don't have time to learn AI, but a lot of them are smart enough to say, hey, we'll go work with Richard.
The and the ones who decide to work with me are the ones who will go above and beyond because it's and it's real simple. Okay? It it goes back to time.
How much time do you have? The guys who say, I'll work with Richard, they don't have to worry about, oh, how am I gonna, you know, see my subcontractors if they if they get back in contact with me?
Oh, how am I going to find out how to get good leads? Oh, how am I going to make sure that everything is scheduled and right each day? All that is ease.
How am I going to advertise? And and when people know about me, you know, if they check my website, what are they gonna see? They don't have to worry about that when they work with me.
I got real proof. And again, if you oh, he's just oh, all he does is advertise you're you're missing the point.
The owner of Maisons Construction does not have to worry about a website. He does not have to worry about inquiries. He does not have to worry about showing his work off in a way that people can see it.
He's not coming home to his computer having to do all this. I took care of it for him. And what you don't realize who's watching this video and saying, oh, man, all Richard does is advertise.
What you don't realize is that you doing all this complaining, meanwhile, you're the frog getting cooked in the slow cooking pot that just boils a little bit at the time because you cannot see what's going on right in front of you. You can't see it.
You gotta you need to open up your eyes because you need to see what's going on. Hey, man. It's over.
Claude Fable five, you do as much as you can before June 22 rolls around.
I know I am. Hey, man. I got a whole website here that is dedicated to what is your business and the biggest pain you want solved and you can literally oh, what slows you down in construction?
Subs go dark. Boom. You got you got an app right here and you got a premium website showcasing the app ready to buy.
You think I'm out here advertising. I'm not. I'm not.
You're confused. You don't understand what's going on because you think, oh, I'll just vibe code my own construction website. Okay.
Let's see how much it cost. Let's see how much it cost in today's June 9.
Let's see how much it cost in thirteen days.
And in thirteen days, you have to learn as much as you can about AI so that you could make something like this.
Come on, man. Come on, man. I mean, you got thirteen days because OpenAI, all these boys let me tell you let me let me say this last thing.
Okay? The investor money is running dry, IPOs are coming out of this and with IPOs need to come profitability.
If you think OpenAI and Google are not gonna follow suit, you are kidding.
Apple put a whole local model on everyone's computer and is still limiting how much they can use token wise. You didn't buy the hardware, you didn't wanna work with Richard, you didn't wanna listen to Alex Finn and you think, oh, everything is going to go good with me, and you're working as a data analyst for a company, and you don't even know what's going on.
You don't even know that they are plotting on your head right now because they got somebody like me who knows. Hey, man.
I'll do the data analysis for your whole company within ten minutes. Just do the do because I know AI and you don't.
Y'all don't understand, man. It's over. It's done.
June 22, everything ends.
The Hook
The bait, then the rug-pull.
A X post drops cold: Claude Fable 5 just widened the intelligence class divide, subscription limits burn in under an hour, and the productivity gap between those who can pay and those who cannot is about to be dozens of times wider. Richard Echols is already three videos into the day.
Named ideas worth stealing.
08:10list
The Three-Move Hedge Against Pricing Shifts
- Own local hardware capable of running open models
- Build skill depth to know when frontier models are worth the cost
- Redirect entertainment spend toward AI investment before the price floor disappears
The video's implicit prescription for surviving the move to usage-based AI billing.
Steal forAny presentation on future-proofing a business against AI cost inflation
04:25model
The Vendor Ratchet Playbook
- Subsidize frontier model access on flat subscriptions to create dependency
- Hook users into workflows they cannot easily unwind
- Quietly nerf subscription-tier models to push upsell pressure
- Move frontier access to usage-based billing once lock-in is deep enough
How the host frames the AI pricing transition as deliberate vendor strategy rather than market forces.
Steal forExplaining SaaS vendor lock-in cycles to non-technical audiences
How they asked for the click.
VERBAL ASK
17:10product
“I got a whole website here that is dedicated to what is your business and the biggest pain you want solved.”
Repeated at least three times throughout the video with screen shares of rmdw.ai. Framed as proof of the thesis rather than a direct pitch.
MENTIONED ON CAMERA
FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
OTHER LINKSAlso linked in the description.
Visual structure at a glance.
