Artificial Intelligence

[Spekulatius]

On 9/29/2025 at 3:45 PM, Dalal.Holdings said:

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

 

 

 

 

It’s impressive but what do a Joe Schmoe upstart comes along and proves that well we don’t need all that stuff, we invented  XYZ and can do this 100x cheaper and it’s almost as good. It’s not like this never happened before (Deepseek anyone), as well as in other domains (PC versus Mainframes/ Minicomputer).

[Dalal.Holdings]

16 minutes ago, Spekulatius said:

It’s impressive but what do a Joe Schmoe upstart comes along and proves that well we don’t need all that stuff, we invented  XYZ and can do this 100x cheaper and it’s almost as good. It’s not like this never happened before (Deepseek anyone), as well as in other domains (PC versus Mainframes/ Minicomputer).

 

Not only that, wait till mom & pop next door see what their power bill is. The politicians are going to have to put a stop to a lot of this expansion which means some of these data centers could be giant paperweights. 

 

https://www.bloomberg.com/graphics/2025-ai-data-centers-electricity-prices/?accessToken=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzb3VyY2UiOiJTdWJzY3JpYmVyR2lmdGVkQXJ0aWNsZSIsImlhdCI6MTc1OTIyNzkzNCwiZXhwIjoxNzU5ODMyNzM0LCJhcnRpY2xlSWQiOiJUM0RFRzlHUFdEM1EwMCIsImJjb25uZWN0SWQiOiJENjJDNjQ5QzExNTk0MkZGQUZDQjUyMjBFMkEyQjlGQSJ9.uZ_4kiyRpkOXL6xUM3VdVp33r0muP-LSTIzwxHWOp5k

 

Quote

A Bloomberg News analysis of wholesale electricity prices for tens of thousands of locations across the country reveals the effects of the AI boom on the power market with unprecedented granularity. The locations and prices were tracked and aggregated monthly by Grid Status, an energy data analytics platform. Bloomberg analyzed this data in relation to data center locations, from DC Byte, and found that electricity now costs as much as 267% more for a single month than it did five years ago in areas located near significant data center activity.

 

[WayWardCloud]

Re: Meta - Totally agree, if Meta wanted to rationally maximize profit they would act like Apple. They don't have a natural foot in the door in to serve business solutions like Microsoft does or to assist and inform people in their every day life like Alphabet does, so I agree they should only use AI to refine their ads algorithms + offering their users better tools to generate AI videos and improve engagement but they would be totally fine being a year or so behind the bleeding edge, using a "good enough" model that's 20x cheaper. But Zuk has always wanted to leap into new categories so he's thrown money at the newest trends anytime one has showed up in case that's the next big thing. I don't believe he thinks of his company as a social media company but as a tech company who happened to get its head start in social media. The man just has big ambitions.

 

Re: Power - the real problem is politicians not removing the red tape for hyperscalers to be allowed to build their own power generation on site. If they remove those barriers to competition with sleepy local monopolies the tech giants will innovate and make electricity much better, cleaner and cheaper than the current decrepit grid we have, leapfrogging legacy utilities companies just like SpaceX made a laughing stock of NASA.

Edited October 1, 2025 by WayWardCloud

[Fly]

13 hours ago, Spekulatius said:

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.

 

 

 

Hot take: Zuck is a wannabe idiot. 

[rogermunibond]

AI hyperscale power issues are already being figured out.  Some of it is going to be voluntary curtailment and switching to backup generation/batteries/VPP during peak load conditions.  And where it is available, some of it is going to be BYOG (build your own generation).  What won't happen because ratepayers won't tolerate it, is passing on a disproportionate cost of newbuild generation to existing ratepayers to the benefit of hyperscalers.

 

In the PJM region, where there are GW of data centers already, the recent capacity auctions resulted in huge increases.  There's likely going to be a requirement to BYOG and along with it streamlined process for transmission and approvals.

[billybobjovialdechicoutimi]

Maybe it's been discussed before... pardon me if so and thanks in advance for a link to the relevant thread... but

 

How is it that the market is terrified that CSU's business will be crushed by AI, but nobody has such fears for Shopify... (or at least, its stock only gets an erotic caress)

 

Can AI not reproduce in one second Shopify's entire infrastructure?

 

Thanks for any insights, I am not smart enough to get it myself

[Malmqky]

17 minutes ago, billybobjovialdechicoutimi said:

Maybe it's been discussed before... pardon me if so and thanks in advance for a link to the relevant thread... but

 

How is it that the market is terrified that CSU's business will be crushed by AI, but nobody has such fears for Shopify... (or at least, its stock only gets an erotic caress)

 

Can AI not reproduce in one second Shopify's entire infrastructure?

 

Thanks for any insights, I am not smart enough to get it myself

 

Payment processing is tricky. You have to be careful storing customer information, especially credit card information. I would be extremely wary of using AI to spin-up an ecommerce store as an entrepreneur, especially when it does stupid things like expose API keys constantly, etc. As a developer, this actually might be the last thing I'd consider using AI for.

 

Shopify also has great integration with things like email marketing, logistics, etc. It's cheap enough where creating and hosting your own ecommerce store, email server, etc. doesn't save you a ton of money.

 

So that leaves companies like Amazon, Squarespace, etc. that might be able to integrate AI to make it extremely easy for customers to create ecommerce stores, or a new competitor. But at that point, why can't Shopify just leverage the same technology? So I would say in that regard, AI is probably neutral to positive for Shopify.

 

In terms of AI reproducing Shopify in one second...well Shopify has how many thousands of features, integrations with infra, compliance, etc? It's taken thousands of engineers and hundreds of millions of dollars to build. Things aren't so simple where AI can reproduce it in a second.

Edited October 3, 2025 by Malmqky

[Castanza]

3 minutes ago, Malmqky said:

In terms of AI reproducing Shopify in one second...well Shopify has how many thousands of features, integrations with infra, compliance, etc? Good luck.

 

This whole AI boom has definitely glossed over the complexity of software and how it's implemented alongside infrastructure, third party software etc. I'm definitely not up and up on everything, but I work in the tech field and I do not see many opportunities for a push button replacement. 

 

@WayWardCloud, @rogermunibond and a few others are definitely more up to date with this info. Have either of you seen an example of a push button solution at scale? 

[billybobjovialdechicoutimi]

4 hours ago, Malmqky said:

In terms of AI reproducing Shopify in one second...well Shopify has how many thousands of features, integrations with infra, compliance, etc? It's taken thousands of engineers and hundreds of millions of dollars to build. Things aren't so simple where AI can reproduce it in a second.

Yeah, can you not say the same about the number of hours it took to develop all the VMS software owned by CSU? and it's much more granular over there and difficult to reproduce, I would think, no?

 

Keep in mind, I am not saying that Shopify does not have a strong moat, I am discussing the matter on a relative basis, vs. CSU... as in the person panicking about CSU's business should also be panicking about SHOP... now conceivably the names are held by completely different people, but I dont think that's the case

[billybobjovialdechicoutimi]

and then there is also the small matter of Tobias' compensation choices (which as a CSU investor, makes me want to vomit)

[Malmqky]

4 minutes ago, billybobjovialdechicoutimi said:

Yeah, can you not say the same about the number of hours it took to develop all the VMS software owned by CSU? and it's much more granular over there and difficult to reproduce, I would think, no?

 

Keep in mind, I am not saying that Shopify does not have a strong moat, I am discussing the matter on a relative basis, vs. CSU... as in the person panicking about CSU's business should also be panicking about SHOP... now conceivably the names are held by completely different people, but I dont think that's the case

Agreed. Also, I think due to the hardware requirements for many VMS, not to mention institutional knowledge, etc. CSU is also fine. If anything AI might be a small beneficiary for them.

Edited October 3, 2025 by Malmqky

[Spekulatius]

On 9/30/2025 at 10:42 PM, WayWardCloud said:

Re: Meta - Totally agree, if Meta wanted to rationally maximize profit they would act like Apple. They don't have a natural foot in the door in to serve business solutions like Microsoft does or to assist and inform people in their every day life like Alphabet does, so I agree they should only use AI to refine their ads algorithms + offering their users better tools to generate AI videos and improve engagement but they would be totally fine being a year or so behind the bleeding edge, using a "good enough" model that's 20x cheaper. But Zuk has always wanted to leap into new categories so he's thrown money at the newest trends anytime one has showed up in case that's the next big thing. I don't believe he thinks of his company as a social media company but as a tech company who happened to get its head start in social media. The man just has big ambitions.

 

Re: Power - the real problem is politicians not removing the red tape for hyperscalers to be allowed to build their own power generation on site. If they remove those barriers to competition with sleepy local monopolies the tech giants will innovate and make electricity much better, cleaner and cheaper than the current decrepit grid we have, leapfrogging legacy utilities companies just like SpaceX made a laughing stock of NASA.

This is a good take on Meta. On power centers / power generation , I have a differnt take. Short of a new generation of nuclear reactors that is probably decade away from commercialization (yes these things laws take longer than people think, there is no silver bullet out there) Power generation is a mature tech and there are no quantum leaps or Moores law like progress expected..

 

Thats why I think the AI boom will  raise power prices through the roof and the winner will be whoever figures out how run AI with much less power consumption, not who figures out how to build ten power plants cheaper than the other guy.  

Edited October 3, 2025 by Spekulatius

[dealraker]

20 hours ago, Spekulatius said:

This is a good take on Meta. On power centers / power generation , I have a differnt take. Short of a new generation of nuclear reactors that is probably decade away from commercialization (yes these things laws take longer than people think, there is no silver bullet out there) Power generation is a mature tech and there are no quantum leaps or Moores law like progress expected..

 

Thats why I think the AI boom will  raise power prices through the roof and the winner will be whoever figures out how run AI with much less power consumption, not who figures out how to build ten power plants cheaper than the other guy.  

Enjoying this discussion.

[WayWardCloud]

21 hours ago, Spekulatius said:

Short of a new generation of nuclear reactors that is probably decade away from commercialization (yes these things laws take longer than people think, there is no silver bullet out there) Power generation is a mature tech and there are no quantum leaps or Moores law like progress expected..

Thanks for explaining! I'll defer to you as the engineer. The comparison I used was SpaceX versus NASA on purpose because there was no Moore's law and no ground breaking leap there either. Just the unforgiving reality of a heavy industry bound to the limits of physics and launchers that had been supposedly refined over many decades and well over a trillion of public dollars. Yet when the private sector showed up the price per kilogram of payload to orbit got divided by 20 or something. No idea if utilities can be improved by that much or how, again I'm no engineer, but I've learned to never underestimate the difference between sleepy monopolies spending someone else's money and what the top talent in the tech world can make happen if you let them loose. The ability of Elon et al. to hire the absolute best and brightest in the world, put their feet under a flame and demand results yesterday can really move mountains.

Edited October 4, 2025 by WayWardCloud

[MungerWunger]

https://www.wsj.com/tech/ai/openai-amd-announce-massive-computing-deal-marking-new-phase-of-ai-boom-ed92cc42?st=QphtBH

 

Under the terms of the deal, OpenAI committed to purchasing 6 gigawatts worth of AMD’s chips, starting with the MI450 chip next year. The ChatGPT maker will buy the chips either directly or through its cloud computing partners. AMD chief Lisa Su said in an interview Sunday that the deal will result in tens of billions of dollars in new revenue for the chip company over the next half-decade.

 

The two companies didn’t disclose the plan’s expected overall cost, but AMD said it costs tens of billions of dollars per gigawatt of computing capacity.

 

OpenAI will receive warrants for up to 160 million AMD shares, roughly 10% of the chip company, at 1 cent per share, awarded in phases, if OpenAI hits certain milestones for deployment. AMD’s stock price also has to increase for the warrants to be exercised.

Shares

 

 

Edited October 6, 2025 by MungerWunger

[Castanza]

@dealraker words sit heavy with me when I think of AMD and NVDA...Back when I first started investing I had 5k worth of each that would be worth many many multiples of what they are trading at now. I remember dumping both for a double....NVDA $30-60 or something like that and AMD $5-15...Had no idea what I was doing of course (both buying and selling)...

[rogermunibond]

@Castanza haven't seen anything yet that does "push button" automation, but that's not to say lots of investment dollars aren't being thrown at how to do this.

 

there are certain middleware standards that would need to be adopted industry-wide to make this easier.  one of which is MCP (model context protocol).  As these standards codify, we may see more software disruption.

 

https://modelcontextprotocol.io/docs/getting-started/intro

 

A blog post from Steven Sinofsky on the middleware of AI - provides a lot of useful context for how this works and what might happen

 

https://hardcoresoftware.learningbyshipping.com/p/230-mcp-its-hot-but-will-it-win?utm_source=post-email-title&publication_id=82387&post_id=160043222&utm_campaign=email-post-title&isFreemail=true&r=8j4ms&triedRedirect=true

Edited October 6, 2025 by rogermunibond

[Castanza]

22 minutes ago, rogermunibond said:

@Castanza haven't seen anything yet that does "push button" automation, but that's not to say lots of investment dollars aren't being thrown at how to do this.

 

there are certain middleware standards that would need to be adopted industry-wide to make this easier.  one of which is MCP (model context protocol).  As these standards codify, we may see more software disruption.

 

https://modelcontextprotocol.io/docs/getting-started/intro

 

A blog post from Steven Sinofsky on the middleware of AI - provides a lot of useful context for how this works and what might happen

 

https://hardcoresoftware.learningbyshipping.com/p/230-mcp-its-hot-but-will-it-win?utm_source=post-email-title&publication_id=82387&post_id=160043222&utm_campaign=email-post-title&isFreemail=true&r=8j4ms&triedRedirect=true

 

Interesting thanks for sharing

[Jaygo]

1 hour ago, Castanza said:

@dealraker words sit heavy with me when I think of AMD and NVDA...Back when I first started investing I had 5k worth of each that would be worth many many multiples of what they are trading at now. I remember dumping both for a double....NVDA $30-60 or something like that and AMD $5-15...Had no idea what I was doing of course (both buying and selling)...

 

I can do so so much worse. In 2015 I subscribed to a publication called nates notes. I bought 5 or 6 stocks at 5K each from the list including NVDA and a spec pharma play called Mannkind. Anyway after a few months I sold the nvda for a 30% gain and went all in on Mannkind. I rode that down about 80% and basically blew up my TFSA for the second time. I just looked and according to google finance NVDA during 2015-2016 NVDA was under one dollar split adjusted. First time in a long while I thought about that. investing is hard when your dumb and try to gamble. 

[rogermunibond]

I have you all beat on NVDA, having invested in NVDA and ATI Technologies (both GPU makers, ATI was acquired by AMD) back in 2005.

 

Let's say that the returns I got were excellent but the regret is real!

[Jaygo]

2 minutes ago, Jaygo said:

 

I can do so so much worse. In 2015 I subscribed to a publication called nates notes. I bought 5 or 6 stocks at 5K each from the list including NVDA and a spec pharma play called Mannkind. Anyway after a few months I sold the nvda for a 30% gain and went all in on Mannkind. I rode that down about 80% and basically blew up my TFSA for the second time. I just looked and according to google finance NVDA during 2015-2016 NVDA was under one dollar split adjusted. First time in a long while I thought about that. investing is hard when your dumb and try to gamble. 

 

Oh no! i just checked my email history from Nates notes. It was actually 2014 and NVDA was sub $0.50 cents With rough math that 5k would be approximately $ 1.85 million. Just painful!

[Castanza]

I also had 20 BTC in the Spring of 2013 (shoutout to Slush pool and the uni library IT  tech who let us setup a mining pool with the desktops) and sold when they were $50...distinctly remembering buying a ton of wings with them for some bros in the dorms...

 

Maybe I should just stop selling shit...

Edited October 6, 2025 by Castanza

[Libs]

On 10/6/2025 at 9:30 AM, Castanza said:

I also had 20 BTC in the Spring of 2013 (shoutout to Slush pool and the uni library IT  tech who let us setup a mining pool with the desktops) and sold when they were $50...distinctly remembering buying a ton of wings with them for some bros in the dorms...

 

Maybe I should just stop selling shit...

 

Ugh. I felt physical pain reading that.

 

[Dalal.Holdings]

This is becoming quite comical. One day people will ask how everyone could be so stupid, but I guess that day is not today...

 

 

 

https://www.cnbc.com/2025/10/08/nvidia-huang-amd-open-ai.html

 

[Spekulatius]

Sounds a bit like OpenAI wants to be an operating system of sorts:

https://openai.com/index/introducing-apps-in-chatgpt/