Runway Maths for 2026: New Rules, Same Discipline

Runway is usually quoted as one number, but it rests on half a dozen assumptions and moves whenever one of them does. The discipline is to model it that way.

Why the maths changed

Three forces moved at once. Money has cost more since the zero-rate years ended in 2022, so investors weigh runway differently. AI shifted some costs from headcount to usage, which scales with adoption rather than hiring. And enterprise buyers slowed down, stretching the time from first meeting to signed contract, and from contract to cash.

None of these break the old discipline. They just punish a model built on a single, optimistic line. A model with visible assumptions lets you test each one honestly before the cash forces the question.

Three scenarios, every month

Keep all three cases live and give each one a trigger, so the team knows when to switch from one to the next.

• Aggressive: you hit plan, you raise on schedule, you keep hiring
• Base: growth is real but slower, you trim the nice-to-haves
• Survival: revenue stalls, you extend runway hard and protect the core

Common mistake: Updating the model only before a raise. By then your assumptions are months stale and the board can tell.

Model the new cost curves

AI costs behave like usage, not salary. If your product calls a model on every action, your cost of goods moves with adoption, which is exactly when you can least afford a surprise. Put that line in the model on its own, with its own assumption you can stress.

Do the same for the lengthening sales cycle. A deal that used to close in one quarter and now closes in two delays revenue, ties up pipeline and moves the month cash actually lands. Model the timing, not just the total.

Worth doing: Break cost of revenue into its volatile parts. The line that scales with usage deserves its own assumption and its own scenario.

Make the triggers boring and specific

"If net new ARR is below plan for two consecutive months, we move to the base case and pause the two open roles." Specific triggers remove the emotion from hard calls.

If a trigger needs interpreting, it will turn into a debate at the worst possible time.

Once the model is connected to live data, reviewing the triggers each month should be a short check rather than a rebuild.

Bringing the board along

Show the board the curve and the triggers, not a single number you will be forced to walk back. A founder who has already named the survival case gives the board less to worry about than one projecting confidence the data does not support.

Runway maths in 2026 is about making sure no decision catches the team by surprise. Wire the model to live data, name the lines that scale, and let the triggers do the hard part for you.