Because AI will either kill us, or make us rich – and the Israel-Iran War will end sooner rather than later because it’s very unpopular.
intercst
2 Likes
Also because of TINA, there is no alternative?
2 Likes
Stocks have been going up because earnings have been going up at high rates. For example:
Excluding Alphabet and Amazon, the blended earnings growth rate for the S&P 500 for Q2 2026 would fall to 32.0% from 50.4%. However, this would still mark the mark the second consecutive quarter of earnings growth above 25% and the seventh consecutive quarter of double-digit earnings growth for the index.
DB2
3 Likes
For now AI stocks continue to deliver strong earnings as companies continue to invest in data centers. We hope AI makes businesses more productive causing companies to continue AI expansion—maybe for decades.
We have just been through this with EVs where people thought 50% of vehicles sold in 2030 would be EVs. Companies scrambled and invested heavily to build that capacity.
Now it seems that 50% is unlikely in that timeframe. Companies are scaling back and writing off some investments.
The same could happen to AI. We shall see. Most believe there will still be some winners but the trick is to figure out which ones.
It could. At the same time we have just seen “the seventh consecutive quarter of double-digit earnings growth”. That seems like a good reason for stock prices to go up (rather than some gravity defying explanation).
DB2
I don’t think that is quite factual. Most legacy companies (NVDA, GOOG, MSFT, TSMC) have strong earnings but the companies leading the way in AI are largely unprofitable. They have strong revenue but negative or anemic earnings.
A very cool site I found while verifying my data:
6 Likes
If we tried to list all the companies that benefit from AI, I wonder how long the list would be. NVDA is the clear leader. Alphabet is another. I see Microsoft mostly as a cloud company, second after Amazon. Taiwan Semiconductor manufactures for Nvidia and does well but also makes lots of commodity chips. Memory chips seem to be doing well. Hynix, Micron, Samsung. Equipment suppliers like Caterpillar, Generac, Cummins, Corning Glass, GE Vernova, Baker Hughes, etc. The list is long. Yes, ChatGPT and Anthropic are not yet public. Have lots of growing to do. Clearly are well known with consumers giving them a marketing advantage. Oracle.
Yes, manufacturers of stuff are doing well - but the companies that are actually selling AI, not as much. That is the risk. Even legacy companies that are building AI models are getting their profits from their legacy products and services.
I am not convinced that AI services won’t simply become a commodity where the cheaper product is dominant - even at 80% of the productivity.
Is there enough different between Claude, ChatGPT and Kimi K3 (China’s leading AI model) to make it worth paying a premium? I remain a skeptic.
I compare it to the decision most people make when they buy a laptop. Very few non-gamers care whether it comes equipped with NVidia vs AMD - and often it is integrated so even less of a decision.
1 Like
I’ve done really well with Sandisk. Picks and shovels, I guess.
But I agree with you, with this many players who “can’t afford not to be in it”, margins will evaporate to nothing (not that they’re not nothing already). You can have a nice monopoly or duopoly. Once you get 4 or 5 competitors in the ring, prices crater absent some sort of “lock in”, which I haven’t seen yet in the AI space.
1 Like
Good observation! Yes, AI is the thing but the question for investors is who or what is in the position to best monetize AI. For example, there are multitudes of smart phone brands but only two operating systems, iOS and Android.
- For example, you need AI for self driving. Who is in self driving best positioned to lead the market.
- For example, you need AI for humanoid robots. Who is in humanoid robots best positioned to lead the market.
Another good way to look at it! I’m doing OK with datacenter related businesses.
Anyone care to propose more such scenarios? Would be most appreciated!
The Captain
Ready to fall at any time according to Marc Faber
Marc Faber suggests precious metals, but keep them outside the USA to stop Trump stealing them 
Drug discovery seems to be an obvious beneficiary of AI tools. You might explore the pharma/biotech area. (Lilly, of course, has a long profitable runway ahead of it as GLP research proliferates into many areas.)
DB2
3 Likes
I explored pharma/biotech decades ago and it was a bust. Data centers are more in my line of expertise.
The Captain
It will almost definitely help on the “front end” of drug discovery, the research side of it, and the endless running of permutations, etc. However, the total cost of getting a new drug to market has a very large percentage on the “back end”, the testing phases, the regulatory efforts, the marketing efforts, the education, etc. AI will help a little with the back end, but likely won’t revolutionize it nearly as much.
(I could be wrong about this because it’s off the top of my head with no research at all.)
1 Like
This.
The most expensive part of bringing a new drug to market is all the clinical trials. AI can’t really help with that. We (hopefully) are a long way from asking AI to substitute real world testing of the drug on real people.
3 Likes
Actually it is helping with clinical trials
https://www.sciencedirect.com/science/article/pii/S1386505625003582
1 Like
Artificial intelligence in clinical trials: A comprehensive review of opportunities, challenges, and future directions
Olawade et al.
https://www.sciencedirect.com/science/article/pii/S1386505625003582
AI integration accelerates trial timelines by 30–50% and reduces costs by 40%
DB2
1 Like
Don’t make the mistake of conflating “helping” as the same thing as eliminating.
We will still need thousands of individual patients and hundreds of doctors for each clinical trial. AI isn’t going to (and shouldn’t) get rid of that expense.
3 Likes
Ya, not so much.
From the article:
10.1. Pfizer’s REMOTE trial Initiative
Pfizer’s REMOTE (Research on Electronic Monitoring of OAB Treatment Experience) trial represents one of the first fully virtual clinical trials, leveraging AI and digital technologies for comprehensive remote data collection [181,182]. The study enrolled participants across multiple sites and utilised AI-powered mobile applications with machine learning algorithms that analysed patient-reported outcomes in real-time. Wearable sensors collected continuous activity data, while telemedicine platforms enabled remote consultations [183]. Specific AI performance metrics included: 95 % accuracy in detecting treatment adherence patterns, 87 % accuracy in predicting treatment response within 4 weeks, and 92 % sensitivity in identifying potential adverse events through patient-reported symptom patterns [184].
The trial demonstrated the feasibility of AI-driven virtual trials, achieving 95 % patient retention rates whilst reducing trial costs by 40 % compared to traditional site-based studies.
So this is a single example where the trial was non-invasive and it could be performed and measured remotely.
That is clearly not representative of all clinical trials, and may not be even representative for most.
By contrast, if this was a new cancer drug, it would very likely require blood and urine tests - something not performed remotely. Anything requiring lab results is going to require a clinical setting.
2 Likes
Nobody said they would eliminate human trials. It is accelerating the process and “helping” which I think you said “AI can’t help with it”. Clearly, the articles indicate that it is helping accelerate candidate selection, among other benefits.