Investing in A.I.

[FREE866]

This is just the beginning...

of what?

 

[rembrandt]

of what?

We will see.

 

[frayne]

People have speculated about the AI bubble bursting for some time. If this is it (which I doubt), so be it, that's why diversification is key.

 

[target2019]

This is about proprietary models (ChatGPT, etc.) vs. opensource ones (DeepSeek). LLM with high startup costs vs. Inference model.

“When I say open source, I mean open data, open training algorithms, open weights, and open evaluations—maximum visibility into how they're trained and what they're trained on,” Stoica told Decrypt. “Now we’re in a situation where Chinese companies bet on open source, unlike the U.S., and are clearly ahead."

Source: DeepSeek's Open-Source AI Costs NVIDIA $600 Billion, Triggers $1 Trillion Sector Slump - Decrypt

There's obviously a lot more for inquisitive active investors to look into. Just in a short time we see a vastly changing attitude towards AI and how it is defined in the investor's mind.

 

[SnowballCamper]

Maybe this is just an opportunity to buy. If AI is so great, then an open source model that works at fractional cost just means it will be ubiquitous sooner, and even better with the higher performing chips.

 

[audreyh1]

They still were using NVDAs chips.

 

[target2019]

They still were using NVDAs chips.

Who is "they?"

I need some context, please.

 

[Sunset]

I can remember wishing I could invest in IT stocks or funds in 1999, but I was prohibited because of ethics considerations. A year later I was glad.

How prophetic.

 

[audreyh1]

Who is "they?"

I need some context, please.

The Chinese open source AI ChatGPT was still using NVDA chips.

 

[target2019]

The Chinese open source AI ChatGPT was still using NVDA chips.

Ok.

A) OpenAI owns ChatGPT. That is based on a large learning model, which is very costly (processors, and so on) when the company builds it. ChatGPT is proprietary, and for-profit.

B) DeepSeek is one of several inference models. The cost to start is much less than a LLM, because there is no dependence on an LLM. The software is open-source, and not closed source (like ChatGPT).

The distinctions are not trivial to businesses enhancing their IT and BI (business intelligence) capabilities. Upon closer inspection we can estimate how the AI source uses resources.

From what I've read so far, inference has advantages, as well as disadvantages. All servers, all computers use a combination of processors to achieve best performance. The LLM requires vast capacity behind the scene (in the cloud) to provide the output. The inference model begins with a starter kit installed on your local machine(s). There are higher-order kits, which of course require more GPUs. In either case the business builds its intelligence and the content is entirely under control of the business.

At this time both models have advantages and disadvantages. Competition will ensue, no doubt.

For active investors such as yourself, this introduces a significant twist to investing in AI.

 

[Luvtoride]

Interesting fact I just found with this Deepseek AI App that everyone is so hyped up about. I asked it a question about the Super Bowl and I got this surprising answer. Its knowledge base is 6 months old! Really? Do most AI apps run on that type of lag?

 

[FiveDriver]

I tuned into the Bloomberg China show last night. They had an interview with a high-level Finance person at Race Capital, with ties into China Tech World.

That interview is worth a listen this morning -- She says that Deepseek AI is "cheaper" than current US version by ORDERS OF MAGNITUDE.
For Example -- Our current AI cost is $15 Dollars per the industry Metric.
By comparison, Deepseek performs the same Metric at $.01-- one penny !! That's too significant to ignore.

The Chinese New Year starts Wednesday.....The Year of the Snake.


Bloomberg - Are you a robot?, a

 

[audreyh1]

Ok.

A) OpenAI owns ChatGPT. That is based on a large learning model, which is very costly (processors, and so on) when the company builds it. ChatGPT is proprietary, and for-profit.

B) DeepSeek is one of several inference models. The cost to start is much less than a LLM, because there is no dependence on an LLM. The software is open-source, and not closed source (like ChatGPT).

The distinctions are not trivial to businesses enhancing their IT and BI (business intelligence) capabilities. Upon closer inspection we can estimate how the AI source uses resources.

From what I've read so far, inference has advantages, as well as disadvantages. All servers, all computers use a combination of processors to achieve best performance. The LLM requires vast capacity behind the scene (in the cloud) to provide the output. The inference model begins with a starter kit installed on your local machine(s). There are higher-order kits, which of course require more GPUs. In either case the business builds its intelligence and the content is entirely under control of the business.

At this time both models have advantages and disadvantages. Competition will ensue, no doubt.

For active investors such as yourself, this introduces a significant twist to investing in AI.

Thanks. Fortunately I’m not an active investor.

 

[2HOTinPHX]

Jon Stewart joked....imagine China taking something and making it cheaper?

 

[pjigar]

I know nothing about AI but I know history. Every new and ground breaking technology have propelled mankind into better quality of life and prosperity. Think agriculture, machinery, auto, plane, internet, smartphone, etc. I am sure I am missing a few. I think electrification, renewable and AI are current tech which will do the same for us. But it is very difficult (I would argue impossible) to identify winners/losers when you in the mist of a particular tech era. History has shown that a pioneering company may or may not be a leader after the dust is settled. So I try not to predict the future for any technology winners and invest in every company i.e. broad market index fund/ETF. Just my 2c. YMMV.

 

[OldShooter]

... But it is very difficult (I would argue impossible) to identify winners/losers when you in the mist of a particular tech era. History has shown that a pioneering company may or may not be a leader after the dust is settled. So I try not to predict the future for any technology winners ...

Yes. It is also true that during most of these bubbles the excitement causes the stocks to become overpriced, followed by some degree of collapse. I would prefer to stay away, but holding total market funds this is not an option. So I just take the ride. The copilot checklist is a good one to remember: (1) Sit down. (2) Shut up. (3) Hang on.

Here is investing patriarch Ben Graham (in "The Intelligent Investor") on the problem: " ... we hope to implant in the reader a tendency to measure or quantify. For 99 issues out of 100 we could say that at some price they are cheap enough to buy and at some other price they would be so dear that they should be sold. The habit of relating what is paid to what is being offered is an invaluable trait in investment. In an article in a women’s magazine many years ago we advised the readers to buy their stocks as they bought their groceries, not as they bought their perfume. The really dreadful losses of the past few years (and on many similar occasions before) were realized in those common-stock issues where the buyer forgot to ask 'How much?' "

 

[Markola]

Honestly, even though I think AI is going to be amazing and make a lot of money, I'm just sticking with index funds with the hopes that I'll get my "share" of AI profits. I'll not go after AI specifically in an "AI fund" or similar. I'm guessing there will be a whole "sector" type fund eventually for AI and maybe that's what you're asking about.

100%. We got a taste yesterday, albeit with misrepresented facts, of how AIs will disrupt other AIs, unpredictably and instantly. There’s no way I would try to pick one.

 

[PlayinwithFIRE]

Stuff like FSKAX is already pretty invested in AI companies...not targetted ..but you get my drift

 

[target2019]

100%. We got a taste yesterday, albeit with misrepresented facts, of how AIs will disrupt other AIs, unpredictably and instantly. There’s no way I would try to pick one.

If you don't mind, would you please share an article or two with misrepresented facts?

Thanks.

 

[Markola]

See several on Google News. The first articles which spooked the market said Deep Seek cost only $5 million. Later articles showed a much larger investment.

 

[Montecfo]

Perhaps a bank shot investment in energy (particularly nuclear) or data center REITs

And nat gas. Additional nuclear supply will take time to come online. But I do agree for the longer term.

 

[Montecfo]

Investing in single stocks is risky. Most people here are relatively risk-averse and that is why were are here.

No. You have simply been taught and believed a falsehood. Indexes are made of single stocks. So is any well constructed portfolio.

And risk aversion does not help people retire early if that is what you are suggesting.

Now let's talk stocks shall we?

 

[Alex The Great]

See several on Google News. The first articles which spooked the market said Deep Seek cost only $5 million. Later articles showed a much larger investment.

It seems what they talked about first is the cost to train their model. Which is indeed much less than for competitors. Development cost is separate.
Also, OpenAI stated they do have an evidence DeepSeek still used their model for training purpose. It may explain the lower cost.
About investment: there are some ETFs with AI in focus like AIQ for example. But I would rather invest into some total growth ETF for less risk and better diversification.

 

[Koolau]

I tuned into the Bloomberg China show last night. T


The Chinese New Year starts Wednesday.....The Year of the Snake.

More fireworks in the neighborhood! Rats! (er, I mean snakes.)

 

[Koolau]

I know nothing about AI but I know history. Every new and ground breaking technology have propelled mankind into better quality of life and prosperity. Think agriculture, machinery, auto, plane, internet, smartphone, etc. I am sure I am missing a few. I think electrification, renewable and AI are current tech which will do the same for us.

I don't disagree, but I cling to my tag line as well though YMMV.