Most write-ups of an AI engagement quote the price as a single number and move on. That number hides the only question a sponsor actually loses sleep over: not what the work costs, but what speed costs, and when in the calendar the bill for that speed comes due. When we scoped a raster well-log digitization programme for a Texas onshore operator, we put two prices on the table for the same scope of work. One was a standard track: 100,000 EUR, 4 FTE, 32 weeks. The other was an accelerated track: 180,000 EUR, 6 FTE, 16 weeks. The operator chose acceleration. This is the burn ledger of what that choice cost us to staff, month by month, and what we learned about the real economics of buying time.
At a glance
Three numbers frame the trade the operator made.
Accelerated track price
Standard track price
Premium paid for speed
Two prices for one scope
The work itself did not change between the two quotes. Either way we were building VeerNet, our model for reading curves off scanned paper well-logs, and either way the corpus was the same wall of public Texas RRC raster scans plus the operator's own archive. What changed was the shape of the team and the length of the calendar.
The standard track put 4 full-time engineers on the programme for 32 weeks. The accelerated track put 6 on it for 16. That is not a small reshaping. Half the calendar at 1.5 times the headcount means the accelerated track concentrates roughly the same labour into a window half as long, and it carries the coordination overhead of a larger team working in parallel rather than a smaller team working in sequence. The 80,000 EUR gap between the two prices is the premium for that concentration: more people, more parallel workstreams, more of the project manager's time spent keeping six people in lockstep instead of four.
The operator's reasoning for paying it was straightforward. A digitization backlog is only worth clearing if it clears before the decisions that depend on it. Sixteen weeks of standard delay can be the difference between having a queryable archive in time for a drilling-programme review and missing that window entirely. Koroteev and Tekic, surveying AI in oil and gas upstream, frame this as the recurring upstream pattern: the value of an automation is gated less by its accuracy than by whether it lands inside the operational window that needs it.1 Speed, in other words, is not a vanity purchase. It is the difference between an asset that informs a decision and one that documents it after the fact.
The ledger, month by month
Here is what the accelerated track actually burned. October was billed in two biweekly tranches as the team spun up: 38,820 EUR for the first half of the month and 51,220 EUR for the second, the ramp you see as people onboard, environments stand up, and the data-engineering substrate comes online. November settled into a full-month rhythm at 88,400 EUR. December, with all six engineers at full tilt through the heaviest model-training and integration stretch, came in at 111,220 EUR. Stack those four steps and the cumulative burn for the quarter is 289,660 EUR.2
That cumulative figure is the heart of the ledger, because it crosses the 180,000 EUR price line partway through the engagement rather than at the end. The accelerated envelope is consumed before the calendar is. The instrument below makes the crossing literal: each bar adds a real biweekly or monthly burn, the dashed line is the budget envelope you signed for, and the orange step is the month in which cumulative spend overtakes it. Flip the track switch and the line jumps between the two envelopes, so you can read off where each one is crossed.
Switch the exhibit to the standard track and the 100,000 EUR line sits lower still, crossed even sooner against the same burn curve. That is the point the ledger is built to make and the point a single headline price obscures: neither track's quoted envelope is a ceiling on monthly cash out the door. It is the contracted value of the deliverable, and the burn that produces that deliverable runs against the broader project economics, not against the envelope alone. The roughly 2.5M EUR full-project estimate sitting behind these four months is the real denominator. Against it, the quarter's 289,660 EUR is about 12 percent of the programme, and the 80,000 EUR speed premium is a rounding adjustment on a number that size.
What the premium actually buys
It would be a mistake to read the premium as 80,000 EUR for fewer weeks. What the operator bought was the compression of risk, not just of time. The value of a decision-support capability is dominated by when it becomes available to the people making capital decisions, because a recommendation that arrives after the commitment is sunk cannot change the outcome.1 The accelerated track moves the model, the validated curves, and the queryable archive forward by sixteen weeks. Every decision in that recovered window that can now lean on digitized data instead of a wall of scanned paper is a decision improved at the margin, and in an upstream programme those decisions are measured in millions, not in the tens of thousands the premium costs.
There is a second, quieter line item in the premium that the burn ledger surfaces. A six-person team is not a four-person team with two extra hands. It is a different coordination problem. The accelerated track carries more parallelism, which means more interfaces to keep aligned, more review load, and a project manager whose job shifts from sequencing to orchestration. Some of the 80,000 EUR pays for raw additional FTE; the rest pays for the friction of running them concurrently. We priced that honestly rather than pretending six people in sixteen weeks costs the same per head as four in thirty-two.
Before
100,000 EUR · 4 FTE · 32 weeks
Standard track: smaller team working largely in sequence, lower coordination overhead, deliverable lands after roughly eight months
After
180,000 EUR · 6 FTE · 16 weeks
Accelerated track: larger team in parallel, higher coordination load, deliverable lands in roughly four months
80,000 EUR premium = half the calendar plus the cost of running six people concurrently
Reading a burn ledger honestly
The reason we publish the monthly figures rather than a single price is that the monthly figures are the only thing that lets a sponsor reason about cash, not just cost. A 180,000 EUR contract value tells you nothing about whether you owe 38,820 EUR in the first two weeks or 111,220 EUR in the last month. The accelerated track is back-loaded: the burn climbs as the team reaches full velocity through the training and integration phase, so the heaviest month is the last one, not the first. A sponsor planning cash against this engagement needs the curve, not the endpoint.
The same ledger also disciplines the conversation about scope. Because the four months sit inside a roughly 2.5M EUR programme estimate, every request to widen scope can be checked against the burn rate it implies rather than against the contract value it appears to fit under. The envelope is the price of a defined deliverable; the burn rate is the speed at which money becomes that deliverable. Confusing the two is how an accelerated track quietly turns into an over-run, and keeping them separate is most of what good delivery management on a programme this size comes down to.
The economics of buying speed, in one line
The operator paid 80,000 EUR to halve the calendar and bring a digitized archive, a working model, and a validated set of curves forward by four months. Against a programme estimated near 2.5M EUR and against decisions that the digitized data directly informs, that premium is small and the speed is the asset. The burn ledger is what makes the case legible: not a single number to take on faith, but four real steps you can watch cross the line, and a track switch that shows you what the other choice would have cost. Buying speed is rational exactly when the recovered time lands inside a window that matters. For this operator, on this archive, it did.
What the accelerated-track burn ledger shows
- The same scope was quoted two ways: a 100,000 EUR / 4-FTE / 32-week standard track and a 180,000 EUR / 6-FTE / 16-week accelerated track. The 80,000 EUR gap is the premium for concentrating roughly the same labour into half the calendar at 1.5 times the headcount, plus the coordination cost of running six engineers in parallel rather than four in sequence.
- The accelerated track's real monthly burn climbed from 38,820 EUR and 51,220 EUR in October's two biweekly tranches, to 88,400 EUR in November, to 111,220 EUR in December, for a cumulative 289,660 EUR over the quarter. The burn is back-loaded toward the heavy training-and-integration month, so cash planning needs the curve, not the headline price.
- The contracted envelope is the price of a deliverable, not a ceiling on monthly cash; the burn runs against the roughly 2.5M EUR full-project estimate, where the quarter is about 12 percent and the speed premium is a rounding adjustment. Speed is rational when the recovered window lands in time to inform decisions measured in millions, which on this upstream programme it did.
References
Footnotes
-
Koroteev, D. and Tekic, Z. Artificial intelligence in oil and gas upstream: Trends, challenges, and scenarios for the future. Energy and AI, 2021. The framing that an upstream automation's value is gated by whether it lands inside the operational window it serves, and that decision-support value is dominated by when the capability reaches the people making capital commitments, draws on this survey. https://www.sciencedirect.com/journal/energy-and-ai ↩ ↩2
-
The two track quotes (180,000 EUR / 6 FTE / 16 weeks; 100,000 EUR / 4 FTE / 32 weeks), the monthly burn figures (38,820 EUR and 51,220 EUR biweekly in October, 88,400 EUR in November, 111,220 EUR in December) and the roughly 2.5M EUR full-project estimate are the engagement's own accepted-proposal and progress-report records. The client, field, and personnel are anonymised under operator confidentiality. The corpus drew on the public Texas Railroad Commission (Texas RRC) well-log dataset. https://www.rrc.texas.gov/ ↩
