Wall Street is down today because of this speculation about AI's economic effects. It's well worth reading:
https://www.citriniresearch.com/p/2028gic
56 replies (most recent on top)
@24d And whoever builds will not be able to use expensive proprietary licenses.
some people claim all code at Anthropic and OpenAI is already written by AI
https://fortune.com/2026/01/29/100-percent-of-code-at-anthropic-and-openai-is-now-ai-written-boris-cherny-roon/
If SAS doesn’t build it, someone else will.
"To attract new customers, a new product is needed."
That's ridiculous.
It's already been true where i work for a little while and is only accelerating. We are of course not generating code and just letting it run, but certainly using AI to help generate code, doc, ideas, and more. The value and challenge is moving from the "how" of production to the "what". It shouldn't be all doom and gloom. This is still ludicrously inefficient. Executives here and probably everywhere have 100s of KPIs, and do not know how to take action on their endless dashboards and AI outputs. SAS (or anyone here) should be able to help. The issue at the moment is nobody is hiring. Everyone is scared and frozen and unsure what to do. Good luck all.
@238 Of course you cannot buy Claude and get rid of all software that does anything with data. That's not the current situation.
But going forward, for some customers, it could be. The hyperscalers are advancing into SAS's market. They are a well-funded competitive threat.
@22v It isn’t about not liking to hear bad news but that is a delightfully condescending viewpoint.
It has more to do with the assertion that Viya is just an overpriced Claude.
Just buy Claude and you can get rid of all software that does anything with data :)
It is incredibly simplistic and incorrect.
@20c That was @p9's point, ten days ago: "Isn't Viya just an overpriced version of Claude?"
That post attracted attacks and downvotes. Some people don't like to hear bad news.
But the AI hyperscalers can do simple data analysis and generate reports. That's all some customers need.
The competitive threat from AI is like the early days of Microsoft Excel. Originally, that product could not analyze data. Now, it's all some customers need.
@1zk That sounds very harmonious. It also sounds like AI and open source would achieve this. What specifically does SAS bring to the table that customers will be willing to pay licensing fees for? If their model switches to something more unified and low cost their whole model will collapse.
If SAS attracted new customers, it could avoid layoffs.
To attract new customers, a new product is needed. This single product should satisfy the use cases of all existing customers and include all functionality from the two existing offerings (both old-world and new-world) using a unified programming syntax. It should also be cloud native, scalable, quick to spin up, and less expensive than current offerings. The newer generation of R&D and Analytics developers can create this system with AI assistance in a small fraction of the time spent creating the previous two. Do this and growth will surely follow.
@1z0 Wait you mean they can’t just buy Claude to replace SAS? I was being sarcastic…
growth essential, yes, but math does not necessarily math better with green field targets. new use cases and users and groups within those customers seem more "addressable" than adding a "logo". for example, why did sas seemingly fail to capture the new opportunities PLTR seemed to capture with in gov? who and what qualities of brand new customers would be better than those opportunities? apple, amazon, openAI, anthropic, google, meta - anyone who builds their own linux, LLM, analytics languages - are not really addressable in any meaningful way. chasing bad targets because they are "new" would just waste time and money.
Completely agree. SAS is "sticky", difficult to replace, in financial, pharmaceutical, and other heavily regulated industries.
In other industries, SAS is not "sticky", so is vulnerable to replacement by Open Source and AI.
The "sticky" customers ensure that AI will not be immediately fatal to SAS, any more than Open Source was.
But like Open Source, the impact of AI will not be zero.
SAS needs either more customers, or fewer employees.
@1vn A few things I would point out about the Fortune 500.
These are made up of quite a few heavily regulated industries like banks, pharma, etc.. These companies have been around a long time and many are heavily tied into SAS. Many may never leave SAS.
However, a few points should be made.
1) At some point, SAS needs to grow and not just rely on it's existing base if it wants to survive.
2) These companies have big pockets and I imagine are currently actively looking at ways to use AI to replace existing expensive proprietary systems that also can be customized more specifically to their own needs.
3) Not all of these Fortune 500 companies heavily use SAS. Amazon and Apple, for example, barely use any components at all.
@1vn if you are not sure that attracting new customers is the first priority, here is the math one more time:
1) Revenues are flat at $3B annually.
2) Expenses increase with inflation 2-3% annually.
3) Therefore to maintain profits SAS must cut headcount 2-3% annually.
If SAS attracted new customers, it could avoid layoffs.
if you have 90something % of Fortune 500 as customers, not sure the first question should be "how should we make this new thing and go appeal to new customers".
@1ps Probably. You guys set the bar too high for the likes of me.
@1ph Anyone on the platform is "better than you"
That sensible and detailed list of factors applies to existing SAS customers.
Are there any factors to attract new customers who don't already have SAS?
@1pe That was very kind of you to humor Bullet Point Guy. You are a better person than me.
There are plenty of other reasons but I honestly think they are trolling. Why would someone come to anonymous thelayoff.com to ask such questions?
@1nq doubt it's about objective "product" or low price factors (it's never really that simple). various other factors are probably going on such as:
- you have a strong relationship with SAS as one of your main vendors
- you have old school SAS coding folks mainly; they have not retired
- your data outgrew your old SAS deployment
- you want to analyze your "big data" in the cloud
- the return you get from this analysis is very high, and so your budget is not slashed
- you are willing to stick with your vendor and pay licensing costs (versus the costs of switching)
- former SAS users run your analytics and IT divisions
This shouldn't be rocket science. No one can determine a solid reason to purchase Viya instead of utilizing AI and open source tools? Even just a high level bullet point?
@1nm That is amazing internet insight into an anonymous person who you have no knowledge of. Incorrect but fascinating evaluation nonetheless.
They can't answer the question because they don't know the answer.
@1jd Not sure I saw a personal attack in my response.
If you go back and read the post by @1fe you’ll see that their diatribe of questions was not genuine. They didn’t want or expect an answer. They think that by talking a lot they appear smart and that it somehow takes away from my very simple comment about installing Viya in 30 minutes. I never said that it is a highly scalable system capable of supporting 1000 users and 500 terabytes of data. Common sense indicates that is not what I installed in 30 minutes.
If you want to see a personal attack look right after their question rant
“My guess is that you can't answer most of those questions, and may not even have known that some of them were questions.”
@1j6 Personal attacks are not rational arguments. I do not think I know all the answers; that's why I asked the question. I wrote small parts of Viya, so I'd like to see it succeed; I just don't understand what SAS's plan is.
@1j9 Claude does not require that proprietary data be placed in a public cloud.
If customers want an AI front end for simple data analysis, they can use Claude
Yes, but be sure to upload all of the proprietary company data into the AI cloud to get the best analysis possible.
@1hw or just an indication that no one really wants to spend time on the layoff.com answering those types questions for you. Especially when you seem to think you already know the answers.
@1hk No one can do much with that, without knowing the details @1fe describes. As they suggest, it could be a simple default installation, not suitable for customers' problems.
I'm more interested in @p9's question of why anyone would want to install it. If customers want an AI front end for simple data analysis, they can use Claude, and if they want advanced features, they can use R and Python on Spark. Both alternatives are cheaper than Viya.
It's a strong indictment of the product that no one can answer the simple question of what advantage it offers.
@1h5 And will not here. I don’t answer to you :)
I stand by what I said that I installed Viya 4 in thirty minutes.
Do with that what you want.
@1gt Perhaps you "COULD", but you have not.
@1fe Actually I COULD answer every one of those questions in great detail.
You clearly weren’t looking for answers and were instead trying to intimidate with your diatribe. You win. You sounded really smart.
But in any case I still installed Viya 4 in about 30 minutes. Your ranting doesn’t change that.
Of course you did. Viya 3 or Viya 4? Did you use one of the internal tools to "deploy Viya", or did you consume one of the several pre-packaged "bundles" (for lack of a better word) that are available through the Azure Marketplace (and elsewhere)? If the latter, did you accept the "stock" Viya, or did you configure it to actually, you know, do anything? How long did that take? Did you right-size your Viya deployment? For what data? Small business data, or "big" data? How big? Did you stick to a t-shirt size, or did you "dial it in"? How long did that take? How much are you paying for the virtual machines in your deployment? What size and type? Do you even know? Are they optimized for high IOPS, HPC, or ??? Or did you deploy into an already-existing AKS cluster? How much are you paying for that? Was it optimized for high IOPS, HPC, or ??? How long did that take? How long was your "Viya deployment" up and running? How much did you pay for it? Did you use metered billing?
My guess is that you can't answer most of those questions, and may not even have known that some of them were questions.
The only way you can
[install] Viya ... in about 30 minutes
is to use one of the internal tools to deploy Viya, or accept the default, as-sold "offering" which solves only the simplest customer use-case, so the same end of the market that SAS's competitors now own, and not the "power to know" end of the market that has been SAS's traditional feeding ground. Or I guess you can have done it so many times that you know what to do, but that's an advantage that SAS's customers don't have.
“ What advantage is there to purchase Viya -- instead of Claude, or R/Python on Spark?”
Lack of a weird name?
Have you ever actually, you know, installed SAS?
OMFG - this should have been fixed by now. What is going on???
What advantage is there to purchase Viya -- instead of Claude, or R/Python on Spark?
@1eb weird I just installed Viya this morning in about 30 minutes…
You should take it upon yourself to install, run, and study as many applications as you can.
LOL, no. Have you ever actually, you know, installed SAS? Any version of SAS? At a customer site it takes months and sometimes years to get the SAS "offerings" the customer paid for up and running. Not every customer, not every offering, but enough that it's a disincentive to SAS's customers.
The unfortunate thing about working for SAS is that you are rarely exposed to what the products actually do, and how companies use them to produce value.
Siloed.
I thought that changed. You should take it upon yourself to install, run, and study as many applications as you can.
The unfortunate thing about working for SAS is that you are rarely exposed to what the products actually do, and how companies use them to produce value.
My best GUESS about Viya, is it a platform that allows companies to take many variables, build a model from them, and then find which terms in the model are rife for exploiting. Exploiting certain variables saves “crumbs” of revenue for the company, improving the bottom line.
But I made all that up based on my understanding of linear regression models.