UK — The Register
AI’s epic cash call is matched only by its ability to strangle revenue streams
AI2 min read
A columnist writes about an anonymous developer who used AI to copy big programs like Microsoft Office and Adobe Creative Suite. The column argues this could hurt the revenue of big software companies, and that AI companies may also hurt themselves.
What happened
A developer released source code and program files that copy Microsoft Office and Adobe Creative Suite. The copies were made with AI, which studied the programs and rebuilt them. The copies are early, unfinished versions, but they already do a lot. The developer says the work was done in a 'clean-room' way, meaning the builders did not copy the original code directly.
Key facts
- The copied programs are in 'pre-alpha' state, which means they are very early and still buggy.
- The code was written in Rust, a programming language.
- The article notes that software features cannot be copyrighted, and it is unclear whether any law bans this kind of copying.
- Gartner says there is not enough return on investment to drive AI growth.
- A SemiAnalysis podcast predicts total spending (capex) on AI by 2030 could reach $11 trillion.
Why AI makes this easier
AI is now good at reading and analyzing computer code. The columnist says this made it possible to clone large applications quickly. The article also says AI companies were trained on large amounts of other people's work, which the writer calls the biggest theft of intellectual property in history.
Why it matters
If AI can copy big software cheaply, companies that depend on lock-in may lose money. The writer also says AI may hurt the businesses it depends on. For example, AI answers can take ad revenue away from websites that pay for content, and that can reduce the new data AI needs. The article also raises doubts about how AI spending will pay off.
What to watch
- Whether any law is used against AI-made copies of software.
- Whether software companies move more products to cloud-only subscriptions.
- How cheaper Chinese AI models affect the money AI companies earn.
- Whether companies get a real return on the huge spending planned for AI infrastructure.
Glossary
- pre-alpha
- The earliest stage of making software, when it is incomplete and often has many bugs.
- capex
- Capital expenditure: money spent on long-lasting things like buildings, computers, and data centers.
- Rust
- A programming language used to write fast and safe software.
- lock-in
- When customers are stuck using one company's product because switching is hard or costly.