Executives expect 135%. They're budgeting for 10.

Ask a senior executive what a rival that genuinely runs on AI will be worth in three years, and the answer is remarkable: 2.35 times the value of a comparable firm without it, a 135% premium (Harvard Business Review, June 2026). Ask the same executive where their own AI budget is pointed, and nearly all give a different kind of answer: efficiency. Faster drafting, faster summaries, fewer hours per ticket. The expectation is priced on one thing, and the money is spent on another.

Those two answers cannot both survive contact with arithmetic. A 135% premium is a growth number, and growth numbers are not earned by trimming expense lines. The distance between what executives expect AI to be worth and what their budgets are structured to deliver is the growth blindspot, and most mid-market operating plans for 2026 are sitting squarely inside it.

The premium is priced on growth

Valuation logic is unsentimental about where value comes from. A buyer paying 2.35x for an AI-capable firm is not paying for a leaner version of the same revenue; they are paying for a business expected to find customers faster, decide faster, and compound from a different base. Cost savings get booked once and absorbed into next year's plan. Growth shows up in the multiple, year after year.

Yet when researchers asked the same executives about actual deployment, almost all reported investment aimed at the efficiency side, the very survey that produced the 135% expectation (Harvard Business Review, June 2026). The market's mental model and the market's purchase orders point in opposite directions. One of them is going to be wrong, and absent a deliberate decision, the budget usually wins.

The cost-cutting ceiling is arithmetic

Run the math an honest CFO would run the night before a board meeting. Assume half of the cost base is amenable to AI at all, a generous figure for most mid-market organizations once plant, inventory, freight, and rent are stripped out. Assume agentic tooling then cuts those addressable costs by a full 10%, which would be an excellent program by any published benchmark. The total expense impact is about 5% (Harvard Business Review, June 2026).

Flow a 5% expense reduction through a typical earnings model and firm value rises somewhere around 10%. Every assumption in that chain was charitable, the addressable share, the depth of the cut, the speed of capture, and the result still misses the 135% expectation by an order of magnitude. No amount of execution excellence closes that gap, because the gap is structural. The ceiling belongs to the strategy, not the team running it.

Growth closes what cost cannot

The same valuation work shows the other lever. A sustained two-percentage-point lift in organic growth adds roughly 50% to firm value, and a four-point lift more than doubles it (Harvard Business Review, June 2026). Read those numbers against the ceiling: two points of growth is worth about five times the entire theoretical maximum of the cost program.

The asymmetry is the whole argument. Efficiency gains are bounded by the size of the cost base; growth gains are bounded by the size of the market. For a $200M–$1B organization, the cost base is the smaller of those two numbers by a wide margin. Pointing an agentic workforce exclusively at the smaller number is a choice, and it is the choice almost every buyer in the market is currently making, usually without ever writing the ceiling down.

There is also a competitive clock running on the choice. Efficiency tooling is sold to every rival at identical prices, so whatever fraction of the 10% ceiling a firm captures gets competed away into pricing within a procurement cycle or two. A growth advantage built on faster, better-informed decisions is harder to copy, because it lives in the operating model rather than on a software invoice. The 2.35x expectation already assumes that someone in each market will build that advantage (Harvard Business Review, June 2026). The only open question for a given peer set is which firm it will be, and the firms spending exclusively on efficiency have already answered it, in the negative, in their own budget documents.

Pointing the budget at the gap

Closing the blindspot does not mean abandoning efficiency; it means refusing to stop there. Growth-side deployment puts named agents on revenue work: assembling account intelligence before a competitor's analyst has opened a spreadsheet, answering the cross-functional questions that stall deals for days, giving a function head visibility their counterpart at a rival cannot match. In Milton engagements that work is paired to senior people and scoped to functions where the output is a decision, not a document (internal operating record).

The honest framing for a board is a two-line budget. Line one is efficiency, with its ceiling stated out loud, roughly 10% of firm value under assumptions friendlier than reality. Line two is growth, where two sustained points are worth around 50% of firm value, five times line one. Outcomes on the second line are design targets, never guarantees, but they are the only targets arithmetically capable of meeting a 135% expectation. Budgets reveal beliefs. Right now, most budgets reveal a belief in 10%.