Every company planning for 2027 right now is writing fiction about model capabilities, competitor moves, and cost curves that will be rewritten twice before the fiscal year closes. This is not a new complaint about budgets; the accounting literature has documented for decades that they are time-consuming to produce and tilt managers toward cost control rather than value creation. What is new is the speed at which the assumptions underneath a budget now expire. An annual budget was always a bet that the world would hold still for twelve months. AI has shortened the world's holding-still interval to something closer to a quarter.
Ricardo Semler got to this conclusion early and by temperament. At Semco, as he has told it, budgets ran six months, and a five-year plan was treated as a sign of inattention. He built fourteen businesses without venture capital and never planned beyond three years, on the theory that three years is as far as anyone can see. He liked to point out that "career" comes from the track of a cart: a single groove, ruts included. The six-month budget was the operational version of that suspicion: commit only as far as your headlights reach.
What the evidence actually says
The movement that formalized this instinct, Beyond Budgeting, has a founding case and a sobering literature, and both matter. The founding case is Svenska Handelsbanken, which abandoned formal budgets around 1970 under CEO Jan Wallander, replacing them with radical decentralization: each branch a profit center, performance judged relatively rather than against a negotiated annual number. The accounting literature presents it as the standing demonstration that a large organization can run, and arguably run better, without budgets, though as a single-firm case this is descriptive evidence, not proof.
The sobering part comes from the field studies. Østergren and Stensaker, examining a large energy company that dropped budgets, found the essential result: removing the budget does not remove control. New mechanisms emerge, relationships between headquarters and divisions shift, new dependencies form. Abolition is not liberation; it is substitution. And Matějka, Merchant and O'Grady, comparing 80 Beyond Budgeting adopters with 121 non-adopters, found adopters genuinely more decentralized, with flexible resource allocation and relative targets, but also found many struggling to actually let go of the fixed annual budget for decisions and evaluations. Meanwhile Libby and Lindsay's survey of North American practice found most managers consider their budgeting value-added and prefer improving it to abolishing it. The honest summary: shorter, more adaptive planning works, but the annual budget survives because it does jobs nobody assigned a successor.
Why AI tips the balance
Those jobs are worth naming: the budget is a forecast, a contract (targets against which people are judged), and a permission system (who may spend what). AI changes the economics of the first job completely. Continuous, granular forecasting used to be prohibitively expensive, which made the annual estimate a reasonable economy. When re-forecasting costs almost nothing, an annual number is no longer thrift; it is voluntary blindness, a decision to keep steering by a photograph.
But cheap forecasting also raises the stakes on cadence mismatch. A company that re-forecasts monthly while holding targets and spending permissions fixed for twelve months has built an instrument panel that updates in real time, wired to controls that move once a year. The organization sees the curve coming and is contractually forbidden to steer. This is where the six-month cadence earns its keep: it is short enough that assumptions rarely rot before renewal, and long enough that teams are not permanently in planning mode. It forces the contract and permission functions to re-clear at the speed the forecast function now runs.
Semler's version had one more property worth copying: the short horizon was paired with radical transparency, so the plan was a shared instrument rather than a negotiation weapon. A budget renegotiated twice a year behind closed doors just doubles the gaming. Renegotiated in the open, it halves it.
The uncomfortable question
The objection writes itself: markets, boards, and banks want annual numbers. Fine; produce them, as outputs of a rolling process rather than as the process itself. Handelsbanken reported to markets for decades without an internal budget. The real obstacle is not external; it is that the annual cycle protects everyone who benefits from being measured against a stale number, and every organization contains many such people.
So the question for a CEO in an AI economy is not "should we abolish the budget?", the evidence suggests most firms will not, and need not. It is narrower and harder: what is the longest period for which you are willing to lock targets and spending against your own best current forecast? If the honest answer is six months, and increasingly it is, then the annual budget is not a planning tool. It is a tradition, and traditions should at least know that they are traditions.
Sources
- Østergren & Stensaker (2011). Management control without budgets: a field study of Beyond Budgeting in practice. European Accounting Review. Removing budgets does not remove control; new control mechanisms and dependencies emerge. https://www.tandfonline.com/doi/abs/10.1080/09638180903487842
- Matějka, Merchant & O'Grady (2021). An Empirical Investigation of Beyond Budgeting Practices. Journal of Management Accounting Research. 80 adopters vs 121 non-adopters: adopters more decentralized with relative targets, yet many struggle to abandon the fixed annual budget. https://doi.org/10.2308/JMAR-19-010
- Libby & Lindsay (2010). Beyond budgeting or budgeting reconsidered? Management Accounting Research. Most North American managers view budgets as value-added; dominant path is improving, not abolishing. https://bpb-us-w2.wpmucdn.com/u.osu.edu/dist/8/36875/files/2016/12/Libby-Lindsay_budgeting_survey-17bvsph.pdf
- Lindsay & Libby (2007). Svenska Handelsbanken: Controlling a Radically Decentralized Organization without Budgets (teaching case). Handelsbanken dropped budgets ~1970 under Wallander; branches run as profit centers with relative performance. https://www.researchgate.net/publication/247875136_Svenska_Handelsbanken_Controlling_a_Radically_Decentralized_Organization_without_Budgets