Because i have been thinking about this and something doesn't add up. Reason with me… (p.s. I stand to be corrected.) the hyperscalers Microsoft, Google, Amazon, Meta are collectively spending over $700 billion on AI data centres this year. yeah 700 BILLION DOLLARS. NVDA is printing money. AMAT is printing money. the picks and shovels guys are obviously winning. But who else? Do you know of any company outside of semiconductors that has genuinely moved the needle on revenue because of AI. Not "we integrated AI into our workflow". Not "our AI product is coming soon". I would like to actually know if there is a revenue line that changed. Because from my pov the ROI gap is getting harder to ignore. Palantir trades at a price-to-sales ratio of 86. The historical ceiling before things get ugly is 30. NVDA is 7% of the entire S&P 500 on its own. 57% of the economists in a Deutsche Bank survey said that the AI bubble is the single biggest market risk this year. Bigger than tariffs. Bigger than the Middle East. Bigger than the recession. And yet NVDA just did $68 billion in one quarter. The revenue is real. So i can't figure it out. is the money going to flow downstream to actual businesses eventually? or are we building the most expensive infrastructure in history for returns that never fully materialise??
is anyone actually making money from AI or is it just the chip sellers?
byu/Ready_Poem_3580 ininvesting
Posted by Ready_Poem_3580
19 Comments
I’m paying $200/month for a subscription and it’s easily generating me more value than that. I’d probably pay $500+ / month for the services I’m getting.
Context: software
Almost half the mobiles games on IOS and Android are now built entirely using AI by single persons or small dev team. And they are making guac per minute. That’s just one more example.
The people selling the land for those data centers and lenders financing the equipment are also making a lot of money. Everyone else is bleeding.
This is the next step.
I honestly think six months to a year ago when Microsoft license renewals were coming around with big co-pilot charges businesses were scoffing at the idea of paying big money for subscriptions.
I feel like we have now crossed over the turning point. If your business is not paying for an AI subscription for their staff then they are falling behind.
The revenue for the business and retail products is only just taking off now. Seems Anthropic are ahead of the curve but it’s about to get real, especially as businesses see revenue impact from reducing their workforce.
This is the “gaining market share” phase now.
It is once a de facto monopoly on the product has been established due to market dominance, does the true market pricing power comes, and with it, comes massive profits.
to be fair … alot of nvda’s revenues is coming from the AI ” centipede” .. circular investments…
If you can let go of 30% of employees because AI effectively lets top performers pick up the slack, then that’s a huge impact on the bottom line. I imagine some companies, especially software, are aiming for more than 30%.
The KPI for CEOs since last year has been cost per employee.
The potential future revenue is GINORMOUS. If and when I can legit replace each of my employees who sit in front of a PC all day with an AI agent, that agent is worth that person’s salary. Multiply that to every business in the world.
Plus all those laid off people start their own companies and pay for their own AI agents…
Plus I’ll pay even more because the AI agents can run 24/7/365 with higher productivity and efficiency.
The potential future has SSOOOOOO much automated productivity for everyone, and the AI vendors become the pick and shovel for everyone else in the world.
Edit: I’m not saying any of this is certain or moral, just illustrating the perceived future cash flow that justifies the biggest capex in human history.
Young small business owner kinda ya
A friend run a local b2b business and replace all illustrations & ads with ai and customer don’t care
A relative run a dorm & Airbnb in Bali replace all his logo ads sign instructions and contract with AI images, words, and contracts reviews some in foreign languages
For contract he still hire lawyer to do final review but now he don’t have to pay people to draft one for him or pay to type it
It cut his artist, freelance and law commission cost by a lot and customer don’t care small stuff like that
Basically sector where people don’t care AI is used
My surface level understanding is that it’s a race to AGI and they are throwing everything under the bus to get there 1st, because winner takes all (potential responsibility for ending humankind included).
That’s the part I wonder about too. The infrastructure side is clearly making money, but for most AI software companies, the real ROI still feels unclear.
Most businesses dont make money in the beginning and live of funding
People are definitely making money from AI yes. My job, which was highly sought after just a few months ago, is on the verge of obsolescence. I’m being paid to dig my own grave and set up an agentic framework that does all the things I was doing, but much faster and more exhaustively. The price of inference could go x10 and my company would still happily pay for it. I plan to switch career to something artisanal.
If I count as anyone, then me, but otherwise I dunno.
A friend of mine was running a rocket engine test facility (he’s now retired), and he said use of AI was saving them serious money. Obviously a specialized niche, but it’s an example.
The problem was never “we need to build more faster”. The app store is filled with millions of apps. Everyone is rushing to buy a shovel (AI subscription) but they don’t know what to do with it. It makes developers more productive, but if it’s a big company, coding time was never the issue, politics and meetings are. Startups that found PMF will benefit the most of this but that’s about it.
AI ist the new dot-com bubble. And we all now how that ended.
Total domination of tech companies in the long run
One interesting thing I saw the other day was saying that if you automate all the easy parts of a job with AI, then you’re just leaving your employees with the hard stuff. Even if that’s less work, they’ll get burned out more because they dont have those “easy wins” to sort of pick them up throughout the day as they’re just dealing with complex stuff constantly.
That means they’re taking more time off work (either through breaks or just slower working pace), or you’re having to invest more in their wellbeing. So even if you save money initially, AI might not save you anything in the long term.
The large model makers are in theory not far from profitability. I think bounded domains like coding have probably empirically improved productivity in study.
Other than that, no. I think hyperscalers might be well ahead of their skis. And nvidia honestly is manufacturing some of it’s own demand, so that’s probably distorted.