Comment by WarmWash
1 day ago
I think the underestimation is how many people want a personal knowledge worker in their pocket, and are willing to pay ~$65/mo for it.
1 day ago
I think the underestimation is how many people want a personal knowledge worker in their pocket, and are willing to pay ~$65/mo for it.
Personally, I've only encountered any of those people on line, and almost exclusively here on HN.
Most people I've met -- and again, in a pretty darn skewed sample globally -- see $65/mo as a lot of money to spend on technology of any kind and can't think of anything much they need from "a personal knowledge worker in their pocket". I don't know a single person in real life who remains excited about AI at all, and only a few software engineers who feel it'd be worth that much.
Everybody seems to be mostly confident with the "knowledge productivity" in their personal and professional life and a pretty skittish about spending in today's economy. Most would be excited about a magic new robot that affordably saved them from unwanted physical labor and drudgery, but nobody needs much real help making appointments or filling out forms or whatever.
That's not to say I won't be proved wrong some day, with some further innovations in AI products, but global-scale demand isn't waiting for anything that's been released so far.
I've yet to meet a person that fits that description IRL. Admittedly I don't live in the valley but I do work in tech. The only place I see that demand is on hacker news (and I imagine twitter - I'm not on it).
The competitors of $65/mo subscriptions are the free models and services that are good enough. It will only get worse as open models or free tiers catch up. For most people, they just use whatever that's free
Apple TV, Netflix, BritBox and PBS add up to about $45 a month. Most people are gonna judge AI up against what they’re already paying for and the AI model makers simply don’t have a good enough product.
There’s only two things useful to the average person something to help them translate and something to help them write in everyday life.
Something else that might be useful would be local single purpose AI agents who’s remit is to help you with one specific task, but I don’t think that’s what the people building those expensive data centers want to sell to the market.
> Apple TV, Netflix, BritBox and PBS add up to about $45 a month. Most people are gonna judge AI up against what they’re already paying for and the AI model makers simply don’t have a good enough product.
What's the actual TAM for premium tv subscriptions though? Unlike free models which are keep improving, you can't get premium tv for free. Also they are actually competing with the total subscription price a person is willing to pay for a year.
> There’s only two things useful to the average person something to help them translate and something to help them write in everyday life.
Exactly, free models are good enough for these tasks.
I just don't see how b2c is a large enough market to ask average Joe to cough up hundreds per year when there are free stuff everywhere. B2b is another story