AI companies may face pressure not only from rivals at the frontier, but from cheaper models that are capable enough for everyday work. A new essay argues that this “good enough” market could weaken the case for paying top prices for the most advanced systems.
Watch Desk analysis
What happened
In a 29 September essay for Plagiarism Today, the author argues that lower-cost and open-weight models could meet many users’ needs without matching the best-performing systems. The piece also points to companies building their own AI tools rather than relying entirely on outside providers.
As examples, the essay says law firm Morgan & Morgan plans to spend $1 billion over ten years on its in-house AI system, MX2, and may offer it to other firms by the end of 2027. It says Kirkland & Ellis has announced a separate $500 million investment in an internal platform. Those are examples cited by the essay, not independent measures of how well the systems work.
Why it matters
If a cheaper model is good enough for a particular job, buyers may care more about cost, control and where their data goes than about whether it leads a leaderboard. For AI providers built around selling access to frontier models, that could make the market less winner-takes-all than the grandest investment pitches suggest.
The counterpoint is that building and maintaining an in-house system takes money and expertise too. “Good enough” is not a universal setting, and an open model does not make deployment free. Still, the essay’s examples show why some large organisations may see a tailored system as worth the effort.
Our read
This is a useful argument about the economics of AI, not proof that frontier models are about to be displaced. The most interesting question is what buyers value once model quality clears the bar for their task. A leaderboard win may impress; a system that costs less, fits the workflow and keeps sensitive material in-house may win the purchase order.
What to watch
- Whether the law firms’ stated AI investments produce systems used beyond their own organisations.
- How buyers compare the full cost of in-house models with commercial AI services.
- Whether cheaper and open-weight models prove capable enough for more demanding work.
Discussion spark: When an AI model is good enough for the job, should buyers prioritise price and control over the strongest available performance?
Sources and evidence
- The AI Industry’s ‘Good Enough’ Problem – Plagiarism Today (29 September 2026, 23:45 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.