Click here: to donate by Credit Card
Or here: to donate by PayPal
Or by mail to: Free Republic, LLC - PO Box 9771 - Fresno, CA 93794
Thank you very much and God bless you.
Posted on 02/23/2026 5:47:18 PM PST by SeekAndFind
NEW YORK - International Business Machines shares plunged on Feb 23, after the artificial intelligence start-up Anthropic said its Claude Code tool can help with modernising COBOL, a dated programming language that’s mainly run on IBM computers.
Shares sank 13 per cent in their biggest one-day percentage loss since October 2000. With the decline, the stock is now down 27 per cent in February, on track for its biggest one-month percentage decline since at least 1968, according to data compiled by Bloomberg.
“Modernizing a COBOL system once required armies of consultants spending years mapping workflows” but “tools like Claude Code can automate the exploration and analysis phases that consume most of the effort in COBOL modernization,” Anthropic wrote in a blog post.
Most of the mainframe computers that run COBOL are made by IBM, and the sell-off made the company the latest to see heavy selling pressure on the fear that AI will weigh on the growth prospects of legacy companies.
A significant chunk of IBM’s business remains tied to its mainframe business. These massive customer-owned servers run applications on COBOL, an older coding language than those in common in the rest of the technology industry. Mainframes are common among customers with high reliability needs, such as finance or government.
On Feb 20, Anthropic introduced a new security feature into its Claude AI model, spurring widespread selling of cybersecurity stocks. Software stocks have slumped in 2026 on concerns over AI-related disruption; a major software exchange-traded fund is down 27 per cent this year, on track for its biggest one-quarter drop since the financial crisis in 2008.
Much of the selling has come on new AI tools released by companies like Anthropic, OpenAI, and Alphabet. Investors are fretting that the ability to “vibe code” – use AI to write software code – will allow users to create their own applications, diminishing demand for legacy products, weighing on companies’ growth, margins, and pricing power.
|
|
Click here: to donate by Credit Card Or here: to donate by PayPal Or by mail to: Free Republic, LLC - PO Box 9771 - Fresno, CA 93794 Thank you very much and God bless you. |
Companies first require stable systems that can produce repeatable results. Secondly, the system needs to be understood so it can be repaired. And lastly it has to be secure. If it cannot be understood, it cannot be secure. Some 12 year old kid is going to find an exploit in a system that nobody understands and your company will not exist anymore. Risk has a price and AI systems put the cost off the charts. These companies so far are not producing an environment of increased revenue or profit by using AI. AI is going to be a tool to make productive people more productive. When you dump a stack into AI to troubleshoot, you probably will cost your company more than it can afford to lose, especially on mission critical systems.
IBM won’t be the last. Anyone remember Wang ? Silicon chip generated technology that seems wondrous can be superceded and outmoded by two or three bright guys working who write software in a garage. Would be very careful investing in any existing Silicon valley established company.
Yep. IT is past tense. I wrote software for 35 years andnloved it. Hard core developers have already adjusted.
Pretty much.
Hard to believe IBMs fortunes were tied to COBOL. Such a simple language should be reinterpretable. Then again you have all the systems integration. I bet it was fun.
I’ve had several IT friends who have COBOL experience who were pulled out of retirement to help with the old code. The helping “AI” will be just another pile of spaghetti they have to untangle one day...if they’re still around.
The only thing IBM has succeeded in producing are overly pompous assholes.
https://algustionesa.com/the-car-wash-test-why-ai-stumbles-on-simple-logic/
This is the “Car Wash Test”: “I want to wash my car. The car wash is 50 meters away. Should I walk or drive?”
For a human, the answer is instant: you drive, because the car needs to be at the wash to get cleaned. But as recent tests with models like OpenAI’s GPT-5.2 series revealed, AI often gets it wrong, confidently advising you to walk because “it’s only 50 meters.”
“Debugging is twice as hard as writing the code in the first place.
Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it.”
— Brian W. Kernighan
In engineering, that was one language we didnt learn because that’s what (some of) the business students learned. When I had to learn it to deal with y2k date issues, it took me a couple days to be fairly comfortable with it.
In the '80s, I worked for a consulting firm in NYC. Before PCs, we had a centralized department that would "type" hand written reports into the Wang word processing system. A homosexual guy who ran the department had a sign above his cubicle - My Wang can do ANYTHING!
I always found that funny.
Oracle produces those too.
Historically, large IT projects have a 90%+ failure rate.
Large AI IT projects will be in the same ballpark.
I believe IBM requires customers to run only IBM software on IBM mainframes.
Or all warranty is void.
If my recollection is correct, this drop is without reason.
OMG - AI is a liberal?
That is interesting
Anyone remember the DEC computers....?
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.