# IBM stock falls after Anthropic says AI can now modernize old software **URL:** \ **Category:** Uncategorized\ **Created:** [February 26, 2026, 4:37am UTC](https://fortran-lang.discourse.group/t/ibm-stock-falls-after-anthropic-says-ai-can-now-modernize-old-software/10747 "2026-02-26T04:37:35Z")\ **Posts on this page:** 5\ **Page:** 1
**Author:** ![zaikunzhang](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/zaikunzhang/32/785_2.png) [@zaikunzhang](https://fortran-lang.discourse.group/u/zaikunzhang)\ **Post date:** [February 26, 2026, 4:37am UTC](https://fortran-lang.discourse.group/t/ibm-stock-falls-after-anthropic-says-ai-can-now-modernize-old-software/10747/1 "2026-02-26T04:37:35Z")
[https://www.fastcompany.com/91497326/ibm-stock-falls-anthropic-says-ai-can-cobol-ai-coding](https://www.fastcompany.com/91497326/ibm-stock-falls-anthropic-says-ai-can-cobol-ai-coding) ---
**Author:** ![jorgeg](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/jorgeg/32/6835_2.png) [@jorgeg](https://fortran-lang.discourse.group/u/jorgeg)\ **Post date:** [February 26, 2026, 6:28am UTC](https://fortran-lang.discourse.group/t/ibm-stock-falls-after-anthropic-says-ai-can-now-modernize-old-software/10747/2 "2026-02-26T06:28:49Z")
I have been using Claude specifically to modernise some codebases and although good, it hits some limitations rather quickly which, without deep Fortran experience, they would be very difficult to resolve. However, they provide a very nice and stable avenue to develop tools to help modernise a specific codebase. ---
**Author:** ![Beliavsky](https://avatars.discourse-cdn.com/v4/letter/b/ba8739/32.png) [@Beliavsky](https://fortran-lang.discourse.group/u/Beliavsky)\ **Post date:** [February 26, 2026, 2:10pm UTC](https://fortran-lang.discourse.group/t/ibm-stock-falls-after-anthropic-says-ai-can-now-modernize-old-software/10747/4 "2026-02-26T14:10:20Z")
My approach in [Pure-Fortran](https://github.com/Beliavsky/Pure-Fortran) is to tell Codex to write a Python script to change Fortran code in some way, for example add `intent`, indent, wrap long lines, or group declarations. Then when the script produces code I don’t like I tell it what changes I want, and it modifies the Python code. With this approach you teach LLM how to modernize Fortran code through the medium of code in Python (or some other language), and you get replicable results, since the Python scripts are deterministic. ---
**Author:** ![jorgeg](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/jorgeg/32/6835_2.png) [@jorgeg](https://fortran-lang.discourse.group/u/jorgeg)\ **Post date:** [February 27, 2026, 5:45am UTC](https://fortran-lang.discourse.group/t/ibm-stock-falls-after-anthropic-says-ai-can-now-modernize-old-software/10747/5 "2026-02-27T05:45:04Z")
Yeah! that’s the way, then one ends up with a bunch of very nice scripts that are super helpful in general. Then you can also unit test the python to make sure everything always works! ---
**Author:** ![BFelsher](https://avatars.discourse-cdn.com/v4/letter/b/71c47a/32.png) [@BFelsher](https://fortran-lang.discourse.group/u/BFelsher)\ **Post date:** [March 5, 2026, 4:02pm UTC](https://fortran-lang.discourse.group/t/ibm-stock-falls-after-anthropic-says-ai-can-now-modernize-old-software/10747/6 "2026-03-05T16:02:57Z")
Does anyone use PlusFORT anymore? It’s free for personal use and works very well for me. TONS of old F77 code here from early 80’s up to 90’s written to be compiled with MSFORT, then updated to PGI (which has a lot of non-standard fortran stuff in it) and PlusFORT does a really good job at spagging it up to F90 standard. Plus it runs natively in LINUX and WINDOWS. [plusFORT - Get Free Licence Here! - Fortran UK](https://fortran.uk/fortran-analysis-and-refactoring-with-plusfort/plusfort-licence-download/) I try to avoid AI or any tools that do things that I couldn’t do myself (given enough time).