The Capital-Budgeting Case for AI Physical Security: A 2026 ROI Framework on After-Tax Cost, the OBBBA Depreciation Reset, CapEx vs. OpEx, and Payback Period

Why the capital-budgeting case for AI physical security changed in 2025, before the procurement memo was ever written

The capital-budgeting case for AI physical security is now a different calculation than it was eighteen months ago, and the change came from the tax code rather than the camera. When Congress passed the One Big Beautiful Bill Act in July 2025, it restored 100 percent bonus depreciation on a permanent basis and lifted the Section 179 expensing limit to 2.5 million dollars, reversing a phasedown that would have shrunk first-year expensing to 40 percent in 2025 and eliminated it entirely by 2027. For a security director or risk officer building the business case for a real-time detection platform, that reset moves a large share of the deployment cost into a single tax year, compresses the effective payback period, and lowers the after-tax cost of the investment in a way no vendor discount can match. The economics of the decision shifted upstream of the procurement process, and most security buyers have not yet recalculated.

This ROI framework treats an AI physical security deployment the way a chief financial officer treats any other capital project: as a stream of after-tax cash flows discounted to present value, with a depreciation schedule, a recovery period, and a payback threshold that determine whether the project clears the hurdle rate. The point is not to turn a security leader into a tax accountant. It is to give the non-financial buyer the four numbers a finance committee will actually ask about, so the request arrives framed in the language of the people who approve it. The detection performance earns the meeting. The after-tax math wins the budget.

Pretax price is the wrong number to put in the business case

Most security-technology proposals lead with the sticker price, and most finance committees mentally discard it within the first minute, because the pretax cost of a capital asset is rarely what the asset actually costs the organization. A deployment treated as a capital expenditure is depreciated over its recovery life, and every dollar of depreciation is a deduction that reduces taxable income. For a profitable organization paying a 21 percent federal corporate rate, plus state tax, the after-tax cost of a capital purchase is materially lower than its invoice, and the timing of that benefit, not just its size, drives the present-value math. A business case that ignores depreciation is overstating the cost of the project to the very committee it is trying to persuade.

The distinction that organizes the rest of this analysis is the one between capital expenditure and operating expenditure. Hardware that a buyer purchases and owns, such as cameras, edge inference appliances, and on-premises servers, is a capital expenditure recovered through depreciation over time. A subscription to a cloud-based detection service, by contrast, is generally an operating expenditure expensed in full in the period it is incurred. Neither treatment is inherently superior. They produce different cash-flow timing, different balance-sheet effects, and different conversations with finance, and the right answer depends on the organization's tax position, its appetite for capitalized assets, and how it prefers to fund security. The architecture decision and the accounting decision are linked, which is why a buyer evaluating where the model runs, a question IntelliSee examined in its briefing on edge versus cloud AI inference, is also implicitly making a CapEx-versus-OpEx decision.

None of this displaces the operational case for moving from passive recording to real-time detection. That case is established, and IntelliSee has quantified it in its four-variable ROI framework, which models the return on detection-to-response compression directly. The analysis here is the financial companion to that operational model: once a buyer accepts that AI detection is worth funding, the after-tax structure of the purchase determines how much it really costs and how fast it pays back.

The OBBBA depreciation reset, in the terms a security budget actually uses

Two provisions in the 2025 tax law do most of the work in the AI physical security capital-budgeting case, and both took effect for property placed in service in the 2025 tax year. The first is the permanent restoration of 100 percent bonus depreciation. Under the One Big Beautiful Bill Act, qualified property that is acquired and placed in service after January 19, 2025 is eligible for immediate first-year expensing of its full cost, a reversal of the Tax Cuts and Jobs Act phasedown that had already dropped bonus depreciation to 40 percent for 2025 and was set to reach zero in 2027. Tax advisory analyses from BDO and Grant Thornton confirm both the percentage and the permanence, which matters because permanence removes the timing pressure that used to force buyers to rush a deployment into a closing window.

The second provision is the expanded Section 179 election. The IRS rules permit a business to expense, rather than depreciate, the cost of qualifying property up to an annual limit, and the OBBBA raised that limit to 2.5 million dollars for 2025 with a phaseout threshold of 4 million dollars, both indexed for inflation in later years. The detail that matters for security buyers specifically is eligibility: Section 179 expensing is explicitly allowed for security systems installed on nonresidential real property, alongside fire-protection and alarm systems, HVAC, and roofs. A security deployment is not a borderline case for first-year expensing. It is a named, qualifying category.

Where the asset is depreciated rather than expensed, the recovery period comes from the Modified Accelerated Cost Recovery System. Security and surveillance systems are generally classified as five-year property under IRS Publication 946, recovered using the 200 percent declining-balance method with a half-year convention. The practical effect is that even without bonus depreciation or a Section 179 election, a security buyer recovers the bulk of the asset's cost in the first three years rather than spreading it evenly across the asset's service life. Accelerated recovery front-loads the tax benefit, and front-loaded benefits are worth more in present-value terms than the same dollars recovered later. The order of operations also matters: the IRS generally requires a business to apply Section 179 first and then bonus depreciation to whatever basis remains, a sequence that lets a buyer expense up to the full cost of a qualifying deployment in year one.

The four numbers a finance committee will ask for

A capital request that survives a finance review answers four questions in a sequence the committee already uses, and a security leader who arrives with all four converts the conversation from a cost defense into a project evaluation. The first is the after-tax cost, which is the gross outlay reduced by the present value of the tax benefits the deployment generates. As the framework above shows, a 400,000 dollar owned deployment that qualifies for first-year expensing carries an after-tax cost closer to 316,000 dollars for a profitable buyer at the federal rate alone, and lower still once state tax is layered in. The number that belongs in the business case is the after-tax figure, because that is the cash the organization actually parts with.

The second number is the payback period, the time it takes for the cumulative benefit of the system to equal its after-tax cost. Payback is the metric security investments most often struggle with, because the largest benefit, a prevented or mitigated high-consequence incident, is a probabilistic avoided loss rather than a booked revenue line. The honest way to handle this is the counterfactual methodology that boards and insurers increasingly expect, which IntelliSee detailed in its framework on avoided-incident attribution. Lowering the after-tax cost through accelerated depreciation shortens payback directly, because the denominator the benefits must overcome is smaller from day one.

The third number is net present value, the sum of the project's after-tax cash flows discounted at the organization's cost of capital. NPV is where the timing of the depreciation benefit earns its keep: a tax shield realized in year one is worth more than the same shield spread across five years, and the gap widens as the discount rate rises. A deployment that books its full deduction immediately therefore produces a higher NPV than the identical deployment depreciated on a straight line, even though the total dollars deducted are the same. The fourth number is the internal rate of return, the discount rate at which NPV equals zero, which the committee compares against its hurdle rate. These last two metrics are standard capital-budgeting tools, and the reason to name them explicitly is that framing a security request in NPV and IRR terms signals to finance that the proposal was built for their model, not translated into it after the fact.

What the avoided loss looks like at the moment of detection

The entire after-tax model depends on a benefit that is invisible until the instant it is triggered, which is why it helps to anchor the financial abstraction to the operational reality the platform produces. The expected-loss reduction a finance committee discounts is not a slide. It is the compressed interval between a threat entering a camera's field of view and a verified alert reaching the people who can act on it.

Tying the financial model to this moment also disciplines it. The expected-loss reduction a buyer claims should correspond to the event classes the platform can actually detect, under the conditions the site actually presents. A figure measured under ideal lighting and clear sightlines overstates the benefit if the deployment will run in low light, fog, or partial occlusion, a limitation IntelliSee has documented in its analysis of how computer vision handles occlusion, low light, and adversarial conditions. A conservative model uses conservative detection assumptions, and a conservative model is the one that survives scrutiny from a CFO who has seen optimistic security forecasts before.

Owned hardware versus subscription: the structural fork in the model

Because the tax treatment differs so sharply between a capitalized purchase and a subscription, the funding structure is a first-order decision rather than a procurement detail. The table below lays out the two paths side by side, not to declare a winner but to show which questions each path forces a buyer to answer. The accelerated-depreciation advantage applies only to the owned-hardware path, while the subscription path trades that advantage for predictable expensing, lower upfront capital, and the vendor-durability exposure that comes with depending on a service rather than an asset.

The fork also interacts with the total cost of ownership over the full lifecycle, not just the acquisition. A capitalized deployment carries maintenance, model retraining, and integration costs across its recovery life, and a subscription bundles many of those into the recurring fee. IntelliSee mapped the full lifecycle picture in its total cost of ownership report, and the capital-budgeting model should pull its out-year cash flows from that lifecycle view rather than treating the deployment as a one-time purchase that ends at installation.

A worked sequence: how the after-tax model assembles

Walking the four numbers through a single illustrative deployment shows how the pieces connect. Consider a profitable organization deploying a 400,000 dollar owned AI detection layer over its existing cameras, qualifying for first-year expensing, at a combined marginal tax rate of 25 percent and a 9 percent cost of capital. In year one, the organization expenses the full 400,000 dollars, generating a 100,000 dollar tax shield, which brings the after-tax cost of the deployment to 300,000 dollars. That is the figure the payback clock runs against, not the sticker price.

On the benefit side, suppose the deployment produces 90,000 dollars a year in quantifiable operational savings, drawn from reduced contract-guard hours and faster verified response that lowers loss cost, plus a conservatively modeled expected-loss reduction for the high-consequence event class the system addresses. Against a 300,000 dollar after-tax cost, roughly 90,000 dollars of annual benefit yields a payback inside four years and a positive net present value at a 9 percent discount rate, with the first-year tax shield doing much of the work by shrinking the amount the benefits must recover. Run the identical deployment without accelerated depreciation, recovering the cost on a straight line, and both the payback period and the NPV deteriorate even though the total deductions are unchanged, because the benefit arrives later and is discounted harder. The staffing component of the benefit is not hypothetical; IntelliSee modeled the labor economics in its ROI framework for AI-augmented guard operations, and the loss-cost component connects to its work on workers' compensation economics and loss-cost compression.

The sequence is deliberately conservative on benefits and precise on costs, which is the posture a finance committee rewards. It does not assume the system prevents every incident. It assumes the system compresses the detection-to-response interval, that compression has a modeled loss-cost value, and that the tax structure lowers the cost the benefit must clear. The result is a business case built on the committee's own arithmetic.

Where the capital-budgeting case fits the broader buying decision

The after-tax model is one input among several, and it is strongest when it sits inside a complete evaluation rather than standing alone. A buyer who has lowered the after-tax cost through accelerated depreciation has improved the numerator of the decision, but the durability of the vendor, the substantiation of its claims, and the portability of the architecture determine whether the projected cash flows actually materialize over the recovery life. IntelliSee treated that side of the decision in its market analysis of vendor due diligence, and the two analyses are complements: the tax structure tells a buyer what the system costs, while the diligence framework tells a buyer whether the company behind it will be present to deliver the return the model assumes.

For public-sector and grant-funded buyers, the capital-budgeting case takes a different but parallel form, because the relevant question is funding source and appropriation timing rather than after-tax cost. Those buyers should read the depreciation analysis here as the private-sector counterpart to the procurement-vehicle and grant pathways IntelliSee covered in its briefing on federal and state grant funding. In both cases the discipline is the same: identify how the money is recovered or supplied, model the timing, and build the request around the financial mechanism the approving body actually uses. The technology has matured to the point where it works. The remaining advantage goes to the buyer who can speak the language of the budget.

Continue the research

This ROI framework is part of IntelliSee Intelligence's ongoing analysis of the economics of AI physical security. To build a complete business case, pair it with the four-variable ROI framework for the operational return model, the detection-to-response latency economics framework for the loss-cost mechanism, and the total cost of ownership report for the lifecycle cash flows the model depends on. To understand how IntelliSee fits a buyer's existing infrastructure, see how the platform works or request a risk assessment.