Despite acknowledging that LLMs improve my productivity and am happy with how they have positively affected my life - I'm a bit cynical about the trajectory of this "AI bubble" and am looking forward to it popping sooner rather than later.
I just want to get it over with so we can abandon the hype, recover and move onto a less erratic/volatile investment/business landscape.
Sick of the "AI coding interviews" and having to virtue signal that I think AI will usher in economic nirvana.
Was a little bummed that Micron's earnings didn't send the Q's into a spiral after the insanity of the KOSPI the other day.
I'm less and less convinced there's anything to pop. I had a backlog a year+ long for my engineering team in my last company, plus dreams that weren't even worth investing in. Now... that whole year backlog is maybe two months long. And everyone who competes with that has to keep up.
There are two important points to consider though:
On the bubble pop it's the feedback loop of investments to ROI materialising or not which will dictate if this stabilises in the current trajectory or if the bubble pops and massive investments will get written off (whoever is holding the bag can cry a river and won't be made whole).
But on that point for a team delivering value and demonstrating ROI I'm curious to see how many businesses fly through their roadmaps at whatever velocity and quality LLMs can deliver - and how does that translate to business growth.
Are you achieving more in features and accelerating growth to the same 2 months instead of 12+?
Or has the LLM spend just made 12+ months of "cost to build roadmap" into 2 months of staff + LLM costs anyway without profit to match??
Unless "other lines" go up as well, just compressing that increased cost without the benefits in revenue and growth, it'll still fail the ROI to keep paying the LLMs tax for these frontier labs to recoup their investment and then some...
Yes, we can now achieve in two months what we did in 12, all else being equal. If I was willing to experiment with lower quality, I would've said one month.
Do you remember how Uber was a bubble that was going to pop for years?
Why do you think that didn't happen? They're still barely profitable.
This confuses me. I read that they are hesitant to overbuild more production. If they can't drastically increase production then how do they grow revenue further?
1. This AI-driven demand cycle is a temporary blip
2. None of their competitors will make massive investments in capacity.
3. No state actors will do so despite evidence to the contrary.
Many in silicon have gone out of business due to the boom-bust cycles so being a bit conservative is appropriate but at some point not expanding production becomes an existential crisis.
Or ya know enough governments with the power get really angry and start forcing price controls, nationalizing, using their state spy agencies to steal your secrets, or use unofficial back channels to make your life extremely difficult.
I think this is part of the reason why TSMC is doing so well compared to the others, they seem to play this game(invest or not) much better than others.
It is odd actually. They've already ceded the PC market to Chinese brands so they have to be vary careful around the cyclical nature of their single customer (AI data centers)
Despite acknowledging that LLMs improve my productivity and am happy with how they have positively affected my life - I'm a bit cynical about the trajectory of this "AI bubble" and am looking forward to it popping sooner rather than later.
I just want to get it over with so we can abandon the hype, recover and move onto a less erratic/volatile investment/business landscape.
Sick of the "AI coding interviews" and having to virtue signal that I think AI will usher in economic nirvana.
Was a little bummed that Micron's earnings didn't send the Q's into a spiral after the insanity of the KOSPI the other day.
I'm less and less convinced there's anything to pop. I had a backlog a year+ long for my engineering team in my last company, plus dreams that weren't even worth investing in. Now... that whole year backlog is maybe two months long. And everyone who competes with that has to keep up.
There are two important points to consider though:
On the bubble pop it's the feedback loop of investments to ROI materialising or not which will dictate if this stabilises in the current trajectory or if the bubble pops and massive investments will get written off (whoever is holding the bag can cry a river and won't be made whole).
But on that point for a team delivering value and demonstrating ROI I'm curious to see how many businesses fly through their roadmaps at whatever velocity and quality LLMs can deliver - and how does that translate to business growth. Are you achieving more in features and accelerating growth to the same 2 months instead of 12+? Or has the LLM spend just made 12+ months of "cost to build roadmap" into 2 months of staff + LLM costs anyway without profit to match??
Unless "other lines" go up as well, just compressing that increased cost without the benefits in revenue and growth, it'll still fail the ROI to keep paying the LLMs tax for these frontier labs to recoup their investment and then some...
Yes, we can now achieve in two months what we did in 12, all else being equal. If I was willing to experiment with lower quality, I would've said one month.
Do you remember how Uber was a bubble that was going to pop for years?
Why do you think that didn't happen? They're still barely profitable.
They get 12 percent and us ordinary consumers get to pay a lot more, all because of this stupid play by billionaires.
This confuses me. I read that they are hesitant to overbuild more production. If they can't drastically increase production then how do they grow revenue further?
Prices go up.
That's a dangerous game to play.
It requires them to bet that:
1. This AI-driven demand cycle is a temporary blip
2. None of their competitors will make massive investments in capacity.
3. No state actors will do so despite evidence to the contrary.
Many in silicon have gone out of business due to the boom-bust cycles so being a bit conservative is appropriate but at some point not expanding production becomes an existential crisis.
Or ya know enough governments with the power get really angry and start forcing price controls, nationalizing, using their state spy agencies to steal your secrets, or use unofficial back channels to make your life extremely difficult.
I think this is part of the reason why TSMC is doing so well compared to the others, they seem to play this game(invest or not) much better than others.
It is odd actually. They've already ceded the PC market to Chinese brands so they have to be vary careful around the cyclical nature of their single customer (AI data centers)
Because margin is going up. Produce the same amount, make more money. Also don't over invest in capital, so lower capital cost.