Artificial Intelligence

[Peregrine]

2 minutes ago, rkbabang said:

 

 

I'm not too worried about my job as an engineer (IC design), I'm in my early 50s.  If I get to work another 8-10 years I'd be happy.  But what about someone in their early 20s graduating in Engineering or Computer Science today?  What does AI look like 15-45 years from now?  I think I'd be worried if I was younger.

 

 

Young coders straight out of school have always been a cost center for companies, especially the largest ones. But they still hired them to develop them so that at some point they could do higher level work. Unless you think that AI ultimately replaces that higher level work, companies will still need people to do that.

 

Moreover, software engineers do a lot more than just coding. It's figuring out the solution even is, the design, testing, reviews, monitoring, troubleshooting, debugging, etc. etc. 

[Castanza]

Changes in technology are not 1 for 1 in the job market....We have seen this in the past and in general we adapt. Some jobs are lost some are gained but at the end of the day we always seem to have more work. I'd rather be in my 20's and adaptable than in my mid to late 40's with with 20 years of experience and a wall full of IT certifications as a principle Network Engineer or something. 

 

IT and Tech in general has been becoming very siloed the last decade and a half. I think you're already starting to see companies incubate their employees and train them on specific tech stacks. I've had more trainings the last 2 years than I did my first 5...As @Peregrine said, you still will need people at higher levels making decisions or leading the charge. The only way you get that is through development and experience. Talk to people in tech who are near retirement age now. They have been through so many changes. Things that take them 1 day to accomplish now might have taken them months back when they first started. Resources and tools change constantly and AI is just that. Time will tell, if we are all replaced I'll go be an electrician. These AI Electricity usage statistics are insane and borderline impossible. You need to replace like ~5% of Total US electricity usage in 5 years just to accommodate all of these ambitions. That's just the first wave....Will the new shiny tool put a spotlight on problems to be solved and generate adequate returns? or will it end up sitting in the corner as one of the worst investments of all time? 

 

[Hektor]

32 minutes ago, rkbabang said:

But what about someone in their early 20s graduating in Engineering or Computer Science today?  What does AI look like 15-45 years from now?  I think I'd be worried if I was younger.

I think everyone in IT should be worried. AI will replace some and augment others. Everyone will be forced to reskill.

 

https://www.cnbc.com/2025/09/26/accenture-plans-on-exiting-staff-who-cant-be-reskilled-on-ai.html

 

Accenture plans on ‘exiting’ staff who can’t be reskilled on AI amid restructuring strategy

 

Accenture CEO Julie Sweet outlined plans to cut staff who are unable to reskill on artificial intelligence in an earnings call on Thursday. Sweet said in a call Thursday that as advanced AI becomes “a part of everything we do” and the global professional services company continues to invest significantly in the area, it expects employees to “retrain and retool” at scale. “We are investing in upskilling our reinventors, which is our primary strategy,” Sweet said. She explained that the company is “exiting on a compression timeline” people for whom reskilling isn’t a “viable path.”

[rkbabang]

What I'm seeing in the last few posts is both parts of Amara's Law.  I think some people vastly overestimate the effect of AI disruption in the next 5-10 years and others vastly underestimate its disruption in the next 40+ years.  My personal opinion is that there will be a bubble which burst in the next 5-10 years where companies go out of business, many data centers end up abandoned, companies scale way back on their AI investments, and people think AI turned out to be a disappointment.  You'll have luddites saying the equivalent of Krugman's asinine statement that the internet will end up contributing no more than the fax machine.   A.I. will then grow steadily over the next 25-50+ years to disrupt things that even the current bulls didn't expect.

 

[vinod1]

3 hours ago, Peregrine said:

 

I spoke to Microsoft software engineers who have said the complete opposite. The "last mile" work to getting code to actually work properly is still incredibly human-intensive. There is also a whole reddit sub-forum of software engineers working at big tech firms disputing the claims that their CEOs are making.

 

At the very least, these anecdotes suggest that AI isn't gonna render obsolete an entire employment class. Improvements? Sure. 10x improvement? Probably not.

 

Those are the wrong people to talk to! The core engineering software work at Mag7 and the like is vastly different from the business applications that make up 95% of the software work in the economy. 

 

The vast majority of the software development work is not like the work at Mag 7.

 

Talk to the average software engineer at a financial or non-software company, and you get a completely different picture.

 

Vinod

 

 

[nsx5200]

Wasn't going to post this, but seeing the discussion on AI and its impact here, I think these are relevant:

 

https://www.newyorker.com/magazine/2025/09/29/if-ai-can-diagnose-patients-what-are-doctors-for

A more in-depth article on how AI might be used in medical diagnoses situations.

 

"they asked some doctors to read the A.I.’s opinion before they analyzed cases, and told others to give A.I. their working diagnosis and ask for a second opinion. This time, both groups diagnosed patients more accurately than humans alone did. The first group proved faster and more effective at proposing next steps. When the chatbot went second, however, it frequently “disobeyed” an instruction to ignore what the doctors had concluded"

 

"Bresnahan eventually caught the chatbot mixing up dates and hallucinating blood-pressure readings"

 

If we were to apply these findings to the prediction of demise in values investing or in coding (like for CSU), it seems like those predictions are greatly exaggerated.  On the other hand, looking at the experiences of gaming (ex: DoTA, Go), where AI can accumulate multiple lifetimes of learning in simulated environments on models that are more simplistic, AI can easily out-perform humans.

 

So currently, it seems like there are two major classes of problems: one where there aren't a lot of data, relative to the underlying model for AI to latch on, and the other where either there are a lot of data, or can at least can be generated synthetically, such that the underlying model can be emulated correctly.

 

By classifying problems in those two major categories, I think it would be easier to forecast whether it would be easily disrupted by AI or not.  Like I can see low-value coding that a lot of off-shore companies provides being disrupted by AI, or low-value script-based customer service being disrupted by AI.  Whereas higher-level software architecting, or more critical or complex software component remain in the hands of capable engineers.

 

As to the future impact, I can see people that knows how to use AI as a tool will gain a productivity edge over the ones that don't, similar to how Google-fu is a productivity multiplier.  The ones that can maximize new tool can potentially replace/displace many ones that don't.  That's just how it is.  Just like how corporations that used to look for people with skills working with office productivity tools.  Nowadays, it's assumed they can.  If not, Youtube/Google should easily solve that, at least for basic tasks.  As new tools adds more capabilities, everybody have to adapt, or be left behind.  This has deeper implications on society for the ones that aren't willing, or can't adapt, and that contributes to the great political divergence that we see everywhere.

[treasurehunt]

Looks like Alibaba has joined the ranks of the AI maximalists. Here's the transcript (in Chinese) of Eddie Wu's keynote at the Alibaba Cloud Summit earlier this week: https://www.infoq.cn/article/Da7a5caUL75cqvmmZtKg

 

I had ChatGPT translate the transcript for me. Here's the very high level summary:

Thesis: Achieving AGI is now a near-certainty, but it’s only the starting line. The destination is ASI—Artificial Superintelligence. On this path, large models become the operating system, and AI Cloud becomes the next-generation computer.

 

I didn't see any timelines mentioned, but it didn't seem like Eddie Wu was talking about multiple decades to get to ASI. Also, I didn't see any discussion of the dangers of developing ASI.

[Spekulatius]

Believe it or not ORCL is FCF negative by a significant amount and already levered at 4x EBITDA. What could go wrong? This could be a juicy short or candidate for puts some point.

[Red Lion]

On 9/26/2025 at 12:24 PM, rkbabang said:

My personal opinion is that there will be a bubble which burst in the next 5-10 years where companies go out of business, many data centers end up abandoned, companies scale way back on their AI investments, and people think AI turned out to be a disappointment. 

 

This seems like a reasonable prediction, but there are some huge differences between now and the dot com bubble. A lot of these huge investments in data centers and AI are coming from companies that have insane cash balances and high current taxable income. 

 

I'm obviously thinking of GOOGL, META, and AAPL. 

 

That might lead to a bubble in the Capex cycle, but if its well funded through actual cash on hand and investment grade financings, I think this puts us in a very different situation than the Dotcom days since the vast majority of the investment is coming from the bluest chips this time around. 

 

It also seems like even if this huge Capex cycle is a bust from a ROI standpoint, it's likely going to supercharge the infrastructure for AI advancement and research. Just seeing the enormous use applications after a couple years, it's hard to imagine where the limit might be once these systems get scaled up and benefit from the ecosystem getting setup with this huge Capex boom. 

[Red Lion]

21 hours ago, treasurehunt said:

Looks like Alibaba has joined the ranks of the AI maximalists. Here's the transcript (in Chinese) of Eddie Wu's keynote at the Alibaba Cloud Summit earlier this week: https://www.infoq.cn/article/Da7a5caUL75cqvmmZtKg

 

I had ChatGPT translate the transcript for me. Here's the very high level summary:

Thesis: Achieving AGI is now a near-certainty, but it’s only the starting line. The destination is ASI—Artificial Superintelligence. On this path, large models become the operating system, and AI Cloud becomes the next-generation computer.

 

I didn't see any timelines mentioned, but it didn't seem like Eddie Wu was talking about multiple decades to get to ASI. Also, I didn't see any discussion of the dangers of developing ASI.

 

As an outsider this seems like one of the biggest arguments against regulation of AI. 

 

It feels like we are going to enter into an AI arms race with China, and if the west (USA) is hobbled by red tape, it leaves us with the prospect of China taking quantum leaps forward in technology while we fail to keep up. 

 

It seems like ASI presents great potential danger, but even greater if your adversaries have it and you don't. 

[Red Lion]

19 hours ago, Spekulatius said:

Believe it or not ORCL is FCF negative by a significant amount and already levered at 4x EBITDA. What could go wrong? This could be a juicy short or candidate for puts some point.

 

Time to do a fat secondary offering?

[Spekulatius]

1 hour ago, Red Lion said:

 

As an outsider this seems like one of the biggest arguments against regulation of AI. 

 

It feels like we are going to enter into an AI arms race with China, and if the west (USA) is hobbled by red tape, it leaves us with the prospect of China taking quantum leaps forward in technology while we fail to keep up. 

 

It seems like ASI presents great potential danger, but even greater if your adversaries have it and you don't. 

China spending on AI is much lower than what is happening in the US. China spends much more on industries that have to do  with AI (robotics, self driving cars, semiconductors ) specific use cases  than AI itself.

[Dalal.Holdings]

It’s amazing how much money foreigners are putting into the U.S. to fuel the bubble…kind of like Europeans getting in on U.S. railroads in the 1800s blindly…didn’t work out for a lot of them

 

 

 

 

[Valuebo]

Time to start thinking of ways to short this stuff.

[WayWardCloud]

I'm not necessarily disagreeing on the potential for short term over investment and I find it healthy to question the exuberant tone of some techies but mostly it is wild to me that on an investment forum the thread about what's most probably the next big thing after the PC, the internet and the smartphone is 95% negative statements.

 

Fortunes have been made with every new major technology and trend and I'm sure it will be the case once more. What are the 10x+ opportunities that 10 years from now we will wish we had invested in back when we were too busy being the smart but pessimistic people criticizing from the side line and focusing on what may go wrong and which company is overvalued?

Edited September 28, 2025 by WayWardCloud

[Castanza]

43 minutes ago, WayWardCloud said:

I'm not necessarily disagreeing on the potential for short term over investment and I find it healthy to question the exuberant tone of some techies but mostly it is wild to me that on an investment forum the thread about what's most probably the next big thing after the PC, the internet and the smartphone is 95% negative statements.

 

Fortunes have been made with every new major technology and trend and I'm sure it will be the case once more. What are the 10x+ opportunities that 10 years from now we will wish we had invested in back when we were too busy being the smart but pessimistic people criticizing from the side line and focusing on what may go wrong and which company is overvalued?

Because it’s not clear where the fortunes will be made. Most “AI companies” are not profitable. Long-term sure…but people want to see the horse before they place their bets. 

[Dalal.Holdings]

8 hours ago, WayWardCloud said:

I'm not necessarily disagreeing on the potential for short term over investment and I find it healthy to question the exuberant tone of some techies but mostly it is wild to me that on an investment forum the thread about what's most probably the next big thing after the PC, the internet and the smartphone is 95% negative statements.

 

Fortunes have been made with every new major technology and trend and I'm sure it will be the case once more. What are the 10x+ opportunities that 10 years from now we will wish we had invested in back when we were too busy being the smart but pessimistic people criticizing from the side line and focusing on what may go wrong and which company is overvalued?

As i noted: the 1999-2000 internet companies mostly went bust or went nowhere. The future winners of the internet (Goog, Meta, etc) were not even born/public in 2000.

The only real exception is AMZN which suffered a big, long drawdown.

[Vish_ram]

AI is revenue component for some (NVDA ASML …), cost side (LMND DUOL HIMS SPOT NFLX AXON …) for some and both (GOOGL AMZN ORCL MSFT …) for few.

 

Even if AI bubble bursts for those that have it in cost side will tremendously benefit. 

[Castanza]

1 hour ago, Vish_ram said:

AI is revenue component for some (NVDA ASML …), cost side (LMND DUOL HIMS SPOT NFLX AXON …) for some and both (GOOGL AMZN ORCL MSFT …) for few.

 

Even if AI bubble bursts for those that have it in cost side will tremendously benefit. 

GOOGL was a steal at 13x. But the rest of those are either fairly valued, downright expensive or not pure AI plays but beneficiaries. I mean QQQ or the S&P 500 might be the cleanest way to capture AI upside unless we get another GOOGL sell off. Dell could be interesting for a hardware perspective but I haven’t dug in much. 

[vinod1]

1 hour ago, Vish_ram said:

(GOOGL AMZN ORCL MSFT

 

2 hours ago, Dalal.Holdings said:

As i noted: the 1999-2000 internet companies mostly went bust or went nowhere. The future winners of the internet (Goog, Meta, etc) were not even born/public in 2000.

The only real exception is AMZN which suffered a big, long drawdown.

 

I am sure you are aware, but there is something called "Disruptive Innovation" vs "Sustaining Innovation". If there is disruptive innovation, usually the incumbent companies do not want to cannibalize their profits and it leads to innovators overturning the incumbents. If innovation is sustaining, incumbent companies go on to dominate.

 

Internet ("Dot Com") was definitely disruptive. Many think, and I agree, that AI is more sustaining innovation and existing incumbants are going to be dominant. 

 

US tech companies going all out on AI is much more preferable to resting on their laurels, buying back 3% of their stock every year and calling it a day. If google spends $100 billion for next 4 years that is less than 4 years profits. So pay 4 PE less if you think that is all 100% wasted. If they win and win big, say hello $10 trillion.

 

Vinod

Edited September 28, 2025 by vinod1

[Dalal.Holdings]

Quite impressive...just don't ask where the returns are...

 

 

 

 

[WayWardCloud]

I like Alphabet here because they win either way :

- either AI is the next big thing and they are extremely well positioned to profit

- or it's not, they slash capex and simply go back to printing money with Search with the threat of disruption gone

 

It's like when Facebook became Meta and the stock tanked. Zuk was wrong with the metaverse but so what? He tried, it didn't work, he reverted to printing money with Facebook and Instagram, shareholders profited.

 

The incumbents this time around are VERY different than the ones in '99. They can afford to take massive swings and be wrong. They're also hyper aware of the Kodak story so I doubt they would fail in the same way.

[rogermunibond]

Leading tech firms are building physical models of the world to power "Physical AI"

 

https://www.ft.com/content/ae1c6de2-691e-448d-8bf3-8837cc1e6efb?

 

[Eldad]

Interesting that all the moatiest of the asset light companies are having a pretty big AI scare lately. CSU, JKHY, SPGI, TRI, etc. The game is always changing. 
 

Will be interesting if in the future more asset heavy stuff like oil refineries, etc. get a predictability premium. 
 

 

[Spekulatius]

On 9/29/2025 at 4:02 PM, WayWardCloud said:

I like Alphabet here because they win either way :

- either AI is the next big thing and they are extremely well positioned to profit

- or it's not, they slash capex and simply go back to printing money with Search with the threat of disruption gone

 

It's like when Facebook became Meta and the stock tanked. Zuk was wrong with the metaverse but so what? He tried, it didn't work, he reverted to printing money with Facebook and Instagram, shareholders profited.

 

The incumbents this time around are VERY different than the ones in '99. They can afford to take massive swings and be wrong. They're also hyper aware of the Kodak story so I doubt they would fail in the same way.

On Meta - does anyone have any idea why META invests on bleeding edge AI? I get that they need to use AI ( and have done so for a while) but why do they need to spend like a drunken sailor when thru business is really just running social media properties l Could they just do what Apple does and license the best stuff, and then adapt it to their use case?

 

The Metaverse Fiasko could very well repeat itself but on a grander scale.

 

 

Edited September 30, 2025 by Spekulatius