Four of the biggest highway builds now under way in the United States state their costs in four different ways, and only one of them gives a figure to the dollar. Virginia’s Interstate 64 crossing of Hampton Roads is carried at $3,965,451,641 in the annual financial plan its owner issued on 30 September 2025. It is also the smallest of the four. The precision and the size go together: that number is exact because a single contract fixed it, and a single contract could fix it because the work is one procurement rather than a twenty-year sequence of them.
The rest publish something else entirely. Texas gives a program total. Oregon and Washington give a probability range with a most-likely value inside it. Ohio and Kentucky give the price of a phase. Those are four different kinds of quantity, and a list that sorts them by magnitude is performing an arithmetic operation on units that do not match.
The threshold that generates documents, and the ranking that never follows
Federal law does keep something close to a register of megaprojects, though it was written for oversight rather than for readers. Under 23 U.S.C. 106(h), a recipient of federal assistance for a project with an estimated total cost of $500,000,000 or more must file both a project management plan and an annual financial plan, the latter based on detailed estimates of the cost to complete the project and updated every year on assumptions the Secretary of Transportation finds reasonable. Subsection (i) reaches down to $100,000,000 and asks only that the annual plan exist and be produced on request.
The Federal Highway Administration’s guidance for those filings, issued on 18 December 2014, is where the useful conventions live. It expects at least one agency cost estimate review for the whole project within the twelve months before an initial financial plan, describes that review as an unbiased risk-based probabilistic exercise, and directs that the total in the plan be consistent with the review and reflect the 70th percentile cost. Every figure is to be expressed in the dollars of the year each cost will actually be spent.
So the machinery produces dated, defined, confidence-bounded numbers. What it does not produce is comparability. Nothing in the statute or the guidance fixes where a project’s boundary runs, and nothing requires the plan to be published where a member of the public will find it. National rankings of highway megaprojects are assembled from announcements instead, which is how a twenty-year program total ends up listed against a single construction contract.
Houston, where one name covers eighteen years of separate contracts
The North Houston Highway Improvement Project is the largest of the four whose scope is entirely highway work. When the Texas Department of Transportation announced the start of construction on 28 October 2024, it described a $13 billion effort expected to take eighteen years, rebuilding Interstate 45 north of downtown and rerouting it through the city alongside Interstate 10 and US 59. The department’s project pages divide the work into three segments and state the traffic case behind it: average daily volumes in the corridor are projected to rise by as much as 40 percent between 2015 and 2040.
The same announcement priced two actual construction packages. Segment 3B-1, drainage work installing 12-foot box culverts, carried a budget of $121 million and a 2027 completion target. Segment 3B-2, adding mainlanes, frontage roads and bicycle and pedestrian infrastructure, was put at $695.5 million, beginning in January 2025 for completion by 2030.
A later document gives a size to a still larger package. TxDOT’s pre-procurement status summary for Segment 3C-2, dated 19 December 2025, states the number directly: the estimated design-build cost is $2.1 billion including risk-based contingencies and inflation, updated in June 2025 on the final schematic design. The same document confirms design-build delivery. It also notes that state law sets the proposer stipend at a minimum of 0.25 percent of the design-build price and puts the stipend at $5.25 million for this project. That arithmetic lands on the same $2.1 billion. The package is roughly a sixth of the program headline, seventeen times Segment 3B-1’s budget and three times Segment 3B-2’s.
Portland and Vancouver, where the estimate states its own confidence
The Interstate Bridge Replacement program is the clearest published case of an estimate maturing in public. In December 2022 the program put the cost at $6 billion inside a range of $5 billion to $7.5 billion, noting that a 2020 conceptual estimate had topped out at $4.8 billion, and attributing the increase to inflation, workforce shortages and supply chain conditions.
The spring 2026 program update replaced that with a much larger and much more carefully worded figure. Building the full five-mile corridor is estimated at $13.5 billion to $15.2 billion, a range the program describes as carrying a 45 percent to 85 percent probability, with $14.4 billion adopted for financial planning because there is a 70 percent chance the cost lands at or below it. The composition is stated outright: $7.8 billion of base cost, 54 percent of the total, and $6.6 billion of inflation and risk, 45 percent.
The scope behind the headline complicates it further. Extending light rail from Portland to Vancouver with three new stations sits inside the corridor estimate, so the biggest figure in this group is not a highway figure at all. And with committed funding standing near $5.5 billion, the program has defined a smaller core set of projects at $7.65 billion, leaving it with two published costs that are both current and both correct.
In Hampton Roads the ledger runs to the dollar
Virginia’s crossing is the outlier for auditability rather than for size. A federal cost estimate review finalised on 12 December 2018 modelled 38 risk items and produced an overall project cost at the 70th percentile, in year-of-expenditure dollars, of $3,784,000,000. The design-build agreement executed in April 2019 set maximum compensation at $3,299,997,227, a rise of 0.55 percent on the review’s contract figure. Adding a $335,000,000 construction contingency, $90,000,000 of potential completion incentives and a $73,454,414 bridge repair option produces the total the plan carries today, against a data date of 30 June 2025. What that ledger means in practice is taken up in the Hampton Roads case study.
Cincinnati and Covington, where the published total is a phase
The Brent Spence Bridge Corridor supplies the sharpest reminder that a total needs a boundary. The December 2022 announcement of federal grants described a $3.6 billion project after the Federal Highway Administration required a fresh cost analysis. In March 2026 Kentucky put the first portion of the corridor at $4.05 billion, verified by two independent estimators, with the state’s own share at $1.7 billion. The two figures are not a cost increase from one to the other, because the second covers less: the release states that later phases, including rehabilitation of the existing bridge and the remainder of the corridor (eight miles end to end, per the project fact sheet), will be estimated closer to the time of construction. That sequence is examined in the Brent Spence case study.
What sorting them destroys
Ranked by published cost the order runs Portland, Houston, Cincinnati, Hampton Roads. Every step in that order compares unlike things. Portland’s figure is a most-likely value inside a stated range for a multimodal corridor. Houston’s is a program name stretched across eighteen years of procurements yet to be let. Cincinnati’s is the first phase of a corridor. Only Virginia’s is a contracted total, and even that is a maximum compensation with a contingency bolted on rather than a price anyone will pay.
What survives comparison is the cost of the physical thing. Dollars per lane mile, per structure or per bore can be carried between projects with their assumptions attached, which is why unit costs are the durable currency of this subject and are treated on their own terms in the cost of a mile of interstate. The individual project files, read one at a time, are the other half of the method, and they are what the projects and case studies section exists to work through.
For anyone reading a project page, the practical consequence is small and unglamorous. The word next to the number decides what the number means. Program, phase, core set of projects and maximum compensation each mark a different boundary, and two totals drawn at different boundaries can be listed side by side but never ranked.