Genuine question: who is the target customer for these $200+/mo AI subscriptions? They seem targeted to individuals, not business seats, and I’m having a tough time wrapping my head around who is buying these
Comments
Log in with your Bluesky account to leave a comment
In the spirit of genuine-ness, I think it’s targeted at a prosumer segment.
No idea if it’s working, like for my money there’s simply not enough differentiation between the functionality at this level and the baseline consumer SKU but prosumer seems to be the intent.
I tried it for a month when I was generating a bunch of images for a floor plan layout I was working on but abandoned it after got what I wanted.
It’s useful for some really really niche use cases but they will never get enough scale to serve it profitably
The vibe is “fairly well off senior Silicon Valley coder who wants to move into executive positions”, no idea how that tracks with their actual users though
Been seeing more and more devs sign up for all of them, pick one to control them all, then let them run in loops, chucking each version in git until all of the tests pass. Everyone says the same thing: I feel like an army. And then: The sales team approached me about enterprise pricing 🤣
I've been treating o3/4.1 like a junior dev and it keeps exceeding expectations, reading my comments and fixing things I didn't tell it to do. Have spent the day trying out devstral locally and I'm so stoked about having the computer work while I sleep.
"Me and my models" are fixing up a random half-complete browser version of lemmings I found, going to turn it into a weird midi sequencer once the game is complete. https://doublemover.github.io/LemmingsJS-MIDI/ It's almost done
Honestly, I'm not sure either. You can't buy the new AI Pro/Ultra plans for a Google Workspace account. Generative AI in Workspace has privacy protections that don't apply to the consumer free (or paid?) options. The new features shown off will take a while to show up in Workspace, if ever.
Have a friend who's bootstrapping a startup and she's using a big subscription like this and listing it as a business expense. She's the only person working on it because she doesn't want to take VC funding and all that entails.
I think the real question is whether it saves you two hours of work more than you would have already saved with one of the £20 subscriptions. I’d guess there’s heavily diminishing returns.
Possibly. I also work on a team where we're strongly encouraged to git gud at working with coding AIs, and the dynamic there seems to be that once you find one you like, you stick with it. So less like a coke vs Pepsi thing, more like a brand loyalty to cigarettes.
I can only speculate based on the marketing announcing these subscription, and I think the answer is that they themselves do not actually know. It feels like a solution seeking a problem with a price tag looking to keep the AI bubble from bursting...
I don’t know this, but based on what I know about selling software to tech people at businesses (which is a lot), these are business seats, sold directly to the user or manager, and then reimbursed on a credit card.
It’s hard for businesses to buy things, but easy for them to reimburse.
Those guys who buy the latest iPhone on release day and have an Apple watch and a Vision Pro that collects dust and thought NFTs were cool. Folks who use all of the new features regardless of whether they're useful. Folks who thought that Web3 is next because it's 1 higher than Web2.
employees of tech companies using them for work most likely, but strangely most of the $200/mo tools can be built with the $20 subscription tools with minimal effort
My company pays via our individual personal accounts (though it is a 20-30 person startup). They don't generally spring for for $200 plan though. Not necessary.
Marketing/raising awareness to to that workforce and bosses. Not for individuals. The battle is MSFT, AMZN, GOOG for business share, that's where AI revenue will have to come from.
I know off the top of my head that Gusto for example has an “approved vendors” list and just got rid of their wellness stipend to instead move it to an “AI & education” stipend or something which has OAI as an approved vendor (dark, I know).
Wasn't there some noise about individuals needing to submit paperwork including photo ID for these accounts? Presumably they're getting reimbursed or using a company card to pay. Dunno that they're keeping people from sharing or transferring accounts
Price anchoring, I assume, is a big part of it. The lower tiers look like good values in comparison.
It also puts more advanced capabilities into the hands of power users and influencers - who are willing to pay for the best, and will evangelize the products - without losing money on compute.
I think it’s really because they have a critical mass of cultish people who will spend money without actual measurements of increased output like businesses would have.
I think it is aimed at the "It's one banana, Michael, home much could it cost? Ten dollars?" crowd. I mean, I paid for the $40 GPT pro account for a while but yeah, $200 is just too much for me to justify as an individual.
I’m aware of some people running consultancies using them (not necessarily to ‘sell’ AI to clients but using their knowledge of the tools as a lure for business), some small startups to try and reduce some ops costs, and folks who are designing homebrew apps cheaper than a paid developer.
I think they target tech industry employees with the disposable income and who believe they have to be cutting edge to further their careers. I found it just created a huge echo chamber which is why I ultimately left.
It could be something along the lines of 3D TVs. There are no real customers for it, but we spent billions making it so we gotta try to sell it. Big companies being run terribly, a universal constant.
But I use it on occasion to read more about a topic. Just today about an H-bridge and back-EMF, earlier this week more about the pets of Pippi Longstockings. And last week about the history of Legoland Billund.
Yes, that seems like a lot. I strongly suspect that a lot of current LLM use isn't sustainable with subscription fees that recover the inference cost. Maybe freelance coders? But casual users, spammers, or students who want to use it for term papers aren't going to pay that much.
i’m not an expert, but the vibe i get is mostly wealthy tinkerers, obsessives, hype guys, people processing insane amounts of data, doing lots of complex things that require lots of tokens + compute — somewhat similar to the kinds of person who bought the Apple Vision Pro to a degree…
i think a lot of it comes down to the fact that the top tier plans give early access and more usage of the frontier, early release models — some people just want to be first to use a shiny new toy and are willing to pay
People reffered to as "hustlers" people in the west that are convinced they must outcompete everyone around them to accumulate the most wealth. And they think this super premium AI will give them the edge over everyone else.
Educated guess on anecdata from HS educators and TA friends: Even anti-AI people feel a competitive pressure. It doesn't pay to be the only honest person in a *curved* class full of cheaters.
Pro-AI people have the same pressure. If you're already in for $100K, love AI, what's $1600 more to 'win'?
Fintech and LinkedIn-famous companies with recent fundraising rounds that look cool on the website and encourage other companies to buy the “good enough” bulk licenses for 3k employees at $150/m, or a special rate of $99/m because their purchasing dept is just sooo good at negotiating.
i like the open source models, especially small ones like Llama-3-8B, which run fine on my homelab for a lot of the same use cases. i think people should keep trying to push the limits of home inference
I'm not a part of any tpot. i'm just curious. $200 is like a nice dinner out, i can do a couple of those a month? to get future shock on a daily basis?
look buddy, molly asked a good question and seems to be looking for real answers, not shallow one liners meant to get a bunch of “likes” for cheap dopamine
Boosting marketing lies like "the LLM displayed superhuman diagnostic and reasoning abilities" (check dame's profile) is also entirely optional (and disgusting!), yet here we are
People who sit in their parents basements and manage to pay $60 a year to be racists on voice chat on X-Box Live will find a way to scrape together $200 to use something like this to multiply their efforts of harassment of others online.
I'm willing to bet at least one person thought "well this is slightly cheaper than pluralsight and I can talk to it." I hope they're doing better these days, but
Tech executives who can get someone else to pay for it?
Think I've seen a couple, Satya Nadella and Andrej Karparthy, both claim to have a personal "committee" of LLM chatbots that they take advice from and they both probably have Premium+++ subscriptions.
Comments
No idea if it’s working, like for my money there’s simply not enough differentiation between the functionality at this level and the baseline consumer SKU but prosumer seems to be the intent.
It’s useful for some really really niche use cases but they will never get enough scale to serve it profitably
No one in their right mind.
Which doesn't mean I'm going to buy those things, but if that helps to put it in terms of dollars and cents.
It’s hard for businesses to buy things, but easy for them to reimburse.
It also puts more advanced capabilities into the hands of power users and influencers - who are willing to pay for the best, and will evangelize the products - without losing money on compute.
1. People learning in their own time.
2. People going around the business’s procurement processes.
https://www.techwontsave.us/episode/274_how_brainrot_ai_is_upending_the_internet_w_jason_koebler
I'm assuming you refer to Google's AI Pro?
It also includes 30TB of cloud storage and YT premium. So if you need /use that, it's a small added premium.
But I use it on occasion to read more about a topic. Just today about an H-bridge and back-EMF, earlier this week more about the pets of Pippi Longstockings. And last week about the history of Legoland Billund.
Worth a few euros, to me.
Pro-AI people have the same pressure. If you're already in for $100K, love AI, what's $1600 more to 'win'?
I'm suspecting this market will grow for a while
Aside from folks with a real perceived need, some folks have a lot of disposable income
as it is, i still haven’t run out of my ozone quota, although id go for some deep research & o1-pro
As the shine of the party tricks for LLMs fade, the most dedicated fans of simulated human interaction seeking will remain.
Think I've seen a couple, Satya Nadella and Andrej Karparthy, both claim to have a personal "committee" of LLM chatbots that they take advice from and they both probably have Premium+++ subscriptions.