Part of "How It Gets Built" — Buildermuse's series on the machinery behind modern construction.
New York's Second Avenue Subway Phase 1 cost $4.45 billion for 1.8 miles of tunnel and three stations — roughly $2.5 billion per mile, the most expensive subway ever built at the time — while Paris was extending Metro Line 14 for approximately $370 million per mile and Madrid has historically delivered underground metro for under $200 million per mile. The Transit Costs Project at NYU's Marron Institute, which assembled a database of more than 900 transit projects across 59 countries, found the United States pays roughly 2.5 times the global average per mile of tunnel, and the gap against the cheapest competent builders — Spain, Italy, South Korea, the Nordic countries — runs 4x to 6x. Phase 2 of Second Avenue is tracking toward $7.7 billion for 1.5 miles, which would push the record past $5 billion per mile.
This is not a story about corruption or geology. Manhattan schist is good tunneling rock, and the researchers who have audited these projects line by line found the money disappearing into specific, fixable places: oversized stations, layered consultants, risk-loaded procurement, and staffing rules with no European equivalent. For the contractors, engineers, and public agencies reading this, the cost gap is also the opportunity map — because Congress and the FTA are now paying attention, and the firms that can build closer to European numbers will own the next decade of underground work.
The Scoreboard: What a Mile of Tunnel Actually Costs
The Transit Costs Project's headline numbers deserve a plain reading. New York builds subway at $1.5 billion to $2.5 billion per mile and rising. Los Angeles's Purple (D Line) Extension sections have run $800 million to $1 billion per mile. San Francisco's Central Subway came in near $920 million per mile for 1.7 miles. Seattle and Honolulu occupy similar territory when tunneling is involved.
The European comparison set
Against that: Madrid's late-1990s and 2000s metro expansions delivered dozens of miles at $100 million to $200 million per mile in adjusted dollars. Paris's Grand Paris Express — 124 miles of mostly bored tunnel, the largest transit project in the Western world at roughly $42 billion — averages under $350 million per mile including 68 new stations. Milan, Turin, and Naples build for $250 million to $450 million per mile. Even famously expensive London delivered Crossrail — 13 miles of twin-bore tunnel under the densest city in Europe, 10 new stations, $25 billion total — at under $2 billion per mile for a project with roughly triple the tunneling scope of Second Avenue Phase 1. The pattern holds outside subways too: US highway and water tunnels show the same 2x to 4x premium in the Eno Center for Transportation's project database, which compared 61 tunnel projects internationally.
Where the 6x actually accumulates
No single line item explains a 6x gap; the Transit Costs Project's New York case study decomposed it. Stations are the largest driver — on Second Avenue Phase 1, the three station caverns and their finishes consumed roughly 75% of total hard costs, not the tunnel bores. Soft costs (design, program management, construction management) ran 21% of the project against a 5% to 10% norm in comparable countries. Labor costs and staffing levels contributed a factor the researchers estimated at 1.5x to 2x. Procurement and risk pricing added 15% to 25% through contingency and contractor risk premiums. Multiply those factors together and $400 million per mile becomes $2.5 billion.
Follow the Money: Stations, Consultants, Change Orders
America digs cathedrals and calls them stations
Second Avenue's 72nd Street station cavern is roughly three times the volume of comparable new stations in Paris or Copenhagen. US agencies build full-length mezzanines spanning the entire platform, deep mined caverns instead of cheaper cut-and-cover boxes, and back-of-house space European operators handle with a fraction of the footprint. The Transit Costs Project found Italian station construction at Milan's M4 and M5 cost as little as one-sixth of New York equivalents partly because Italy standardizes station design and builds shallow. When 75% of your project cost is stations, a 3x station premium is the cost problem. MTA leadership has acknowledged this: Phase 2 dropped a planned crossover cavern and shortened mezzanines, cutting an estimated $500 million — proof the savings were always available.
The consultant layer cake
US agencies hollowed out in-house engineering over 40 years and now buy it back at markup. On major American tunnel projects it is routine to find a general engineering consultant, a program management consultant, a construction management consultant, and separate design consultants — each with staff billed at 2.5x to 3.5x salary multipliers, each reviewing the others. The Transit Costs Project documented soft costs of 21% on Second Avenue Phase 1 versus 7% in Madrid and 8% in Istanbul, where agencies keep hundreds of engineers on public payroll. The Eno Center's recommendation is blunt: rebuild owner-side technical capacity, because an agency that cannot check designs itself pays consultants twice — once to design and once to be supervised.
Change orders and the risk auction
American design-bid-build contracts hand contractors 100% completed drawings that are wrong often enough that change orders routinely add 20% to 30% to contract value, litigated one at a time. Worse, US contracts push nearly all geotechnical and third-party risk onto the contractor, who rationally prices it: bidders on New York work have told researchers they add 15% to 25% to bids purely for MTA-specific risk, low-bid selection then punishes honesty, and disputes finish the job. Compare Paris, where the owner retains ground risk, uses a geotechnical baseline report as the shared contractual reference, and pays for actual conditions encountered — bids come in lean because contractors are not underwriting the unknowable.
The Machine Itself Is the Cheap Part
Here is the irony: the tunnel boring machine — the most visibly impressive element — is a minor cost line. A large-diameter TBM costs $30 million to $60 million to purchase, roughly 2% of a multi-billion-dollar project.
How a TBM works, in one paragraph
A TBM is a 400-foot rolling factory. A rotating cutterhead — 20 to 60 feet in diameter, studded with disc cutters — grinds the face while hydraulic thrust cylinders push off the concrete segments behind it with 10,000 tons or more of force. Excavated muck exits by screw conveyor and belt; behind the cutterhead, an erector arm places precast concrete segment rings, typically 5 to 6 feet wide, seven segments per ring, and the machine advances one ring at a time. A well-run earth-pressure-balance machine in decent ground mines 50 to 100 feet per day; Madrid's contractors have averaged over 90 feet per day sustained, while US urban drives frequently average under 40.
Utilization is where TBMs bleed money
A TBM crew and its support train cost $100,000 to $200,000 per day whether the machine advances or sits. US machines sit more: utilization (time actually mining) on American drives commonly runs 30% to 40% against 50%-plus on the best European projects, lost to segment logistics, work-rule breaks, utility surprises, and permitting holds. The launch infrastructure compounds it — a launch shaft or portal box with slurry treatment plant, gantry cranes, and muck handling runs $30 million to $80 million before the machine turns, which is why one long drive beats two short ones and why Grand Paris Express ran up to 21 TBMs simultaneously to amortize fixed costs across 124 miles. For a project-level view of who is digging where in the US right now, see our tracker of 14 active TBM drives across the country.
Staffing: 25 on the machine versus 12
The Transit Costs Project's most cited single finding: New York tunnel projects staffed TBMs with roughly 25 workers per shift where Paris, Madrid, and Istanbul run 9 to 15 for the same machine. Some of that is written into local agreements — hoist operators required for automated elevators, standby personnel, craft jurisdiction lines that put four trades on a task Europe gives to one. Researchers attributed a 1.5x to 2x labor factor to staffing levels and work rules, distinct from wage rates themselves; European sandhogs are well paid too, but a French tunnel crew's total labor hours per foot of tunnel run less than half of New York's. This is a negotiated-structure problem, not a wage problem — which means it is solvable at the bargaining table, as Los Angeles has partially shown with project labor agreements that consolidated classifications.
What Is Being Tried, and What Is Under Construction Now
The federal government is the largest single funder of US tunnels through the FTA's Capital Investment Grants program — $23 billion authorized over five years by the Infrastructure Investment and Jobs Act, with roughly $4 billion appropriated annually. CIG's structure historically rewarded cost inflation (bigger project, bigger grant); the FTA has since added cost-review gates, and GAO's audits of CIG projects have pushed the agency toward earlier risk assessments and independent cost estimate reviews before Full Funding Grant Agreements are signed.
The reform agenda
The Eno Center's 2023 tunnel-cost study and the Transit Costs Project converge on the same prescriptions: standardize station and system designs instead of bespoke-everything; grow owner engineering staff and shrink the consultant stack toward a 10% soft-cost ceiling; adopt collaborative procurement (design-build with early contractor involvement, or construction-manager/general-contractor) with owner-held ground risk and geotechnical baseline reports; size stations to ridership rather than monument ambitions; and publish itemized cost data so agencies can be benchmarked. None of this is speculative — every element is standard practice somewhere that builds for a third of US prices.
The active board
The pipeline that will test these reforms is already under contract. The Gateway Program's Hudson Tunnel Project — $16.1 billion, the largest CIG commitment ever at $6.88 billion federal, twin 2.4-mile bores under the Hudson — has all major tunneling packages awarded and TBMs turning; our full breakdown is at Hudson Tunnel project cost, contractors, and timeline. Second Avenue Phase 2 ($7.7 billion) is in early construction with a $1.97 billion tunneling contract let in 2025. Los Angeles's D Line Section 3 ($3.6 billion) is finishing bores toward Westwood. The Hampton Roads Bridge-Tunnel expansion ($3.9 billion) completed its marine bores in Virginia. Add water and conveyance work — Ship Canal in Seattle, the $9 billion-class Delta Conveyance fight in California, Fort Wayne and Cleveland CSO tunnels — and US underground construction is a $40 billion-plus active market. Subcontracting and supplier opportunities on federally funded pieces of that market surface continuously on our bids board.
Frequently Asked Questions
How much does US tunnel construction cost per mile?
For urban transit tunnels including stations: $800 million to $2.5 billion per mile in recent US projects, with New York at the top of the range and Second Avenue Phase 2 tracking past $5 billion per mile. Highway and water tunnels run lower — roughly $400 million to $1 billion per mile — but still carry a 2x to 4x premium over international peers in the Eno Center's 61-project comparison.
Why are American tunnels so much more expensive than European ones?
Four compounding factors, per Transit Costs Project research: oversized mined stations (75% of Second Avenue Phase 1 hard costs), soft costs of 21% versus a 5% to 10% international norm, TBM staffing at roughly twice European levels under local work rules, and procurement that pushes ground risk onto contractors who price it back at 15% to 25%. Geology, safety standards, and wage rates explain far less than commonly assumed.
What does a tunnel boring machine cost to buy and run?
Purchase price for a large-diameter machine runs $30 million to $60 million, but the operating burn is the real number: $100,000 to $200,000 per day for crew and support train, plus $30 million to $80 million in launch shaft and site infrastructure before mining starts. That is why utilization — 30% to 40% typical in the US versus 50%-plus on top European drives — moves total cost more than the machine's sticker price.
Is design-build cheaper than design-bid-build for tunnels?
Not automatically. Design-build compresses schedule and reduces owner-contractor finger-pointing, but researchers found US design-build tunnels still price heavy risk premiums when owners transfer geotechnical risk wholesale. The models that measurably save money pair collaborative delivery with owner-held ground risk and a geotechnical baseline report — the Grand Paris Express and Scandinavian approach — rather than any particular contract label.
Who pays for major US tunnel projects?
The FTA's Capital Investment Grants program is the anchor for transit tunnels — $23 billion over five years under IIJA, typically covering 25% to 50% of project cost, with the Hudson Tunnel's $6.88 billion the largest single grant. The balance comes from state DOTs, regional sales-tax measures like LA's Measure M ($120 billion over 40 years), toll revenue on highway tunnels, and ratepayers on water tunnels.
Which US tunnel projects are hiring contractors right now?
The Gateway Hudson Tunnel ($16.1 billion) and Second Avenue Phase 2 ($7.7 billion) have major packages awarded with subcontract tiers still forming; LA Metro's D Line and Sepulveda corridor, Seattle's Ship Canal follow-on work, and a deep bench of CSO and conveyance tunnels round out a $40 billion-plus active market. Federal solicitations and subcontracting opportunities tied to these programs appear on our bids board as they post.
Your Action Item for This Week
Spend 30 minutes with the Transit Costs Project's free database at the NYU Marron Institute site — filter to the six most recent US projects and their six cheapest European counterparts, and note the soft-cost percentage and station configuration for each. If you sell into public agencies, that one-page comparison is the sharpest credibility document you can bring to a capital-program meeting this year: agencies are under GAO and FTA pressure to explain their numbers, the reform prescriptions are public, and the contractors and consultants who show up fluent in why the 6x gap exists — and how Madrid closed it — are the ones who will be at the table when the next $16 billion program gets carved into contracts.
More from the How It Gets Built desk
Edited by Ray Kowalski, Methods & Machines desk. New teardowns weekly.
