AI vs. software companies

[73 Reds]

1 hour ago, investmd said:

I'm presuming the AI narrative is the cause for drop in price of software companies ytd? 

 

Large cap companies like Microsoft, Salesforce, Adobe, Intuit all down approximately 25% in first 5 weeks of 2026. How come this board isn't writing more about buying at these prices? Pick individual stocks or buy a basket of software stocks that have been beaten down. Is there a separate thread on topic of blood bath in software stocks? 

Candidly would be a bit disappointed if Berkshire did not initiate a position in MSFT.  Hard to find a better manager than Satya Nadella and there should no longer be any perceived conflict of interest with Gates and his prior relationship with Buffett.

[Malmqky]

10 minutes ago, 73 Reds said:

Candidly would be a bit disappointed if Berkshire did not initiate a position in MSFT.  Hard to find a better manager than Satya Nadella and there should no longer be any perceived conflict of interest with Gates and his prior relationship with Buffett.

+1

[UK]

35 minutes ago, 73 Reds said:

Candidly would be a bit disappointed if Berkshire did not initiate a position in MSFT.  Hard to find a better manager than Satya Nadella and there should no longer be any perceived conflict of interest with Gates and his prior relationship with Buffett.

Good manager would not solve MSFT risks with its software and Open AI exposure. I am not saying these ar sure thing risks, but I personally would be disappointed, if they bought MSFT, instead of adding to GOOGL.

[Spekulatius]

 MSFT also has a risk of getting disrupted. Lot’s of their software isn’t really that good. Software like SharePoint to me looks like it’s ripe for disruption. THeir OpenAI partnership has yielded very little other the lousy Copilot.  At least their cloud business is doing very well , but that’s also tied to OpenAI.

 

I think MSFT and Google have high disruption risk but they are leading the charge to disrupt themselves. This is preferable to being passive but not without risk either.

Edited February 8 by Spekulatius

[Eng12345]

6 hours ago, Spekulatius said:

 MSFT also has a risk of getting disrupted. Lot’s of their software isn’t really that good. Software like SharePoint to me looks like it’s ripe for disruption. THeir OpenAI partnership has yielded very little other the lousy Copilot.  At least their cloud business is doing very well , but that’s also tied to OpenAI.

 

I think MSFT and Google have high disruption risk but they are leading the charge to disrupt themselves. This is preferable to being passive but not without risk either.

Hasn't sharepoint already in essence been disrupted by onedrive? 

[Spekulatius]

1 hour ago, Eng12345 said:

Hasn't sharepoint already in essence been disrupted by onedrive? 

Sharepoint is used for business processes similar to Lotus Notes, which was actually better at this sort of things. So the answer is no. Onedrive and sharepoint are different and there isn’t much overlap.

Edited February 9 by Spekulatius

[UK]

Lots on software:

 

[MungerWunger]

https://www.techinvestments.io/p/the-bull-case-for-software-asml-credo

[Milu]

7 hours ago, UK said:

Lots on software:

 

I only heard of this guy a couple of weeks ago but starting to like him as he seems to have a very similar viewpoint on bitcoin and SAAS as I do. 

[rogermunibond]

MSFT has a huge moat in terms of Active Directory and cybersecurity.  We have all kinds of third party apps with possible AI/LLM use cases but who's going to allow any of that to integrate with the enterprise AD and exchange servers?  For large enterprises this is a huge risk.

[Spooky]

2 hours ago, rogermunibond said:

MSFT has a huge moat in terms of Active Directory and cybersecurity.  We have all kinds of third party apps with possible AI/LLM use cases but who's going to allow any of that to integrate with the enterprise AD and exchange servers?  For large enterprises this is a huge risk.

 

Agreed, I think Microsoft and Amazon are going to be the middlemen through which enterprises access the LLMs (and Google to a lesser extent) since they have the enterprise cloud networks already. At work we are using Microsoft Foundry and can just add LLMs directly into our Azure environment with one click.

 

[tnathan]

On 2/8/2026 at 11:53 AM, Spekulatius said:

 MSFT also has a risk of getting disrupted. Lot’s of their software isn’t really that good. Software like SharePoint to me looks like it’s ripe for disruption. THeir OpenAI partnership has yielded very little other the lousy Copilot.  At least their cloud business is doing very well , but that’s also tied to OpenAI.

 

I think MSFT and Google have high disruption risk but they are leading the charge to disrupt themselves. This is preferable to being passive but not without risk either.

I think Google is doing a better job than MSFT. Have yet to see MSFT rip off the bandaid imo

[Saluki]

Thinking though it I'm convinced that software companies with moats will benefit from AI, while others will be hurt by it. MSFT has a moat with Outlook. Google Docs is free and very good, but offices still pay for MSFT Outlook because it would look weird/stingy/risky if they didn't. AWS is very sticky because cloud is so hard to switch from. Amazon is the #1 company in cloud, but it took them 10 years to migrate off of Oracle because they had it early on. Even though they help customers switch to AWS, it's hard. So Google and other in cloud seem okay. 

 

Companies like Verra mobility that have a lot of software, but also government contracts that take years to work the RFP and bidding process should do well too. It will reduce their costs, but not make it any easier for a guy in his garage to disrupt them. So maybe Palantir  with it's Pentagon contracts is okay.

 

What about TableTrac (TBTC). Their software isn't the moat. They need approval from each state gaming commission for their software to be used in a casino in that state. That's a moat. It's software has never been hacked because the base layer is COBOL which is ancient, and not a lot of people can program it, let alone hack it. So maybe AI will help them because it will allow people to make tweaks in COBOL with AI helping them, even if they are junior software people. COBOL programmers are old and the ones who are still alive charge $500 an hour. 

 

What about other physical stuff like drug development that can be helped with AI? Align Systems (Invisalign) has a huge moat. They give the software free to dental schools and the students train on it, so they use it when they start Dentisting (is that a word?). Maybe AI helps them do the software cheaper, while keeping their moat in place. What about smart meters and the grid? Won't that help? It would be hard to make new power grid, but with distributed power, AI can help.  

 

What about Lexis/Nexis and Westlaw that, like Invisalign, is given free to law students so they use it in practice when they graduate. Can't you reduce the costs to update the info and the libraries, while still keeping the law firms paying for it? They charge $500 a year (last I checked) per user but that's one billable hour for an Attorney. WHat about Bloomberg? Those terminals can be updated quicker. 

[formthirteen]

11 minutes ago, Saluki said:

Thinking though it I'm convinced that software companies with moats will benefit from AI, while others will be hurt by it. MSFT has a moat with Outlook. Google Docs is free and very good, but offices still pay for MSFT Outlook because it would look weird/stingy/risky if they didn't. AWS is very sticky because cloud is so hard to switch from. Amazon is the #1 company in cloud, but it took them 10 years to migrate off of Oracle because they had it early on. Even though they help customers switch to AWS, it's hard. So Google and other in cloud seem okay. 

 

Mag7, at least Google and Meta, has already been disrupted (capex). Role playing with AI got me to this future scenario:

 

Quote

 

Google doesn't die suddenly. It follows the IBM/Microsoft playbook of slow relevance erosion:

1. AI assistants capture 20-30% of high-value queries by 2027
2. Regulatory remedies remove default search protections by 2028
3. Ad revenue peaks and begins declining in real terms by 2027-2028
4. Google Cloud grows but can't compensate
5. The company remains large and profitable but loses its position as the organizing layer of the internet

The single highest-leverage factor: Apple's decision. If Apple ships a deeply integrated AI assistant that replaces Google as the default intent layer on 1.5 billion devices, that alone could trigger the cascade. The probability of Apple doing some version of this is very high — the question is degree and timeline.

 

 

TINA > AI sentiment.

Edited February 9 by formthirteen

[Spooky]

31 minutes ago, formthirteen said:

 

Mag7, at least Google and Meta, has already been disrupted (capex). Role playing with AI got me to this future scenario:

 

 

TINA > AI sentiment.

 

This seems plausible to me. Google Search was one of, if not the best, business models in history. They are pivoting to a new, untested business model with significant capex spending (Meta, Microsoft, Amazon and Google are going to spend 2.1% of total US GDP on CapEx this year). At the margins, the new business is not as attractive as the previous one yet it is being priced at 30x earnings (above the last 5 year average of 24.8x) due to being perceived as the AI winner. With a PEG of 0.7 the market is baking in 40-42% earnings growth. Seems like it could get whacked if we see some margin compression.

[Spekulatius]

On 2/9/2026 at 5:24 PM, formthirteen said:

 

Mag7, at least Google and Meta, has already been disrupted (capex). Role playing with AI got me to this future scenario:

 

 

TINA > AI sentiment.

Yes, Google had 90% market share ins search so if the game changes into AI agents do g research and buying for you, it’s unlikely that Google will remain at 90% - their market share has nowhere to go but down.

 

Same with NS Office. Perhaps the future office is an AI assistant helping with writing or doing PP presentation and that seamlessly not with two different apps. That the office bundle would look pretty vulnerable if not obsolete. Of course MSFT could do this as well, but will their market share remain where it is? Probably not.

 

Thats why the Mag7 racing as fast as they can on this. They know they are gettin disrupted too, so they are playing offense, even though they risk getting low return on their spent if it does not work out or their timing is off.

 

Even with NVDA - the moat is only partly in the hardware but mostly in the CUDA software. What if their software layer is easier to replicate on differnt hardware due to AI toolkits ? Then NVDA moat would be gone. It’s the same thing that happened with Cisco by the way. They had software that was supposedly the moat but it dint hold up because the market got commoditized.

Edited February 11 by Spekulatius

[moatrep]

Verisk tics all the boxes of the second post on page 1.

 

Also I wonder if there is any company that would benefit from more freelancing in programing. Maybe a webpage that conects freelancers with companies or any company that can use freelancers in the most optimun way.

Edited February 10 by moatrep

[MungerWunger]

FWIW, I just received an email from the company I work at saying that we're switching from self-hosted data center instances of Jira to Atlassian's cloud based solution.

 

The deadline for the completion is June 2027 ... 

[Longnose]

On 2/7/2026 at 5:27 PM, Junior R said:

Imagine being a big ecom business that had no vendor support to manage systems that are created inhouse by AI

 

umm... this already is Amazon.  If you are big enough you may have an internal person to escalate to but that internal amazon rep is submitting a ticket to some internal software.  Anyone who has ever sold on Amazon you need to operate your business as if you will never have a human to support you. 

[Castanza]

On 2/7/2026 at 11:27 AM, Spekulatius said:

 Mechanical or electrical engineering may be better than computer science, especially those that know the nuts and bolts on a practical level. Analog stuff > digital. 

+1 

 

Mastering and understanding the basics of anything analog will pay dividends. Even things as simple as learning to read a book. The attention span of kids these days is terrible. Engineers that don’t understand the basics of how systems work or why they work are less effective at solving problems. No amount of AI basic task solving will make them better at that. To a degree Education has lost the Apprentice -> Journeyman model and AI has reinforced it. I think it’s a net negative.  

[Castanza]

14 hours ago, Longnose said:

 

umm... this already is Amazon.  If you are big enough you may have an internal person to escalate to but that internal amazon rep is submitting a ticket to some internal software.  Anyone who has ever sold on Amazon you need to operate your business as if you will never have a human to support you. 

LOL I’ve seen so many customers try to switch from private internal cloud to Amazon or Azure only to find out where they land on the support priority list. “But it’s cheaper!” I try to tell them this in advance, but they always decide to waste 6-12 months on proposals only to scrap them. 

[Castanza]

On 2/9/2026 at 1:35 PM, rogermunibond said:

MSFT has a huge moat in terms of Active Directory and cybersecurity.  We have all kinds of third party apps with possible AI/LLM use cases but who's going to allow any of that to integrate with the enterprise AD and exchange servers?  For large enterprises this is a huge risk.

One massive F#@K up might mark the turn of this AI vs SaaS market. 
 

These massive enterprise systems are often one big mess of legacy and modern software/hardware all  shoe stringed together and bubble wrapped together. I don’t see how AI parses these conglomerates and just “vibes a solution.”

[gfp]

Another -20% day for a software co - market is getting twitchy

 

https://www.cnbc.com/2026/02/11/dassault-systmes-share-price-earnings-software-sell-off.html

[rogermunibond]

Can AI/LLMs reduce the number of CAD seats?  Thinking ADSK etc.

[frommi]

Of course, who needs CAD when the robot builds what you want from a prompt?