Industries / Finance and business services / Finance and banking / Equipment leasing and factoring
Leases and equipment loans
This is the work of paying for a machine, truck or aircraft so that a business can use it without buying it outright: by a lease, in which the financier owns the equipment and rents it out for a set term, or by a loan secured on the equipment itself.
6
Milestones achieved, from 1842 to 2015.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1842
A trust holds the cars
The Philadelphia and Reading Rail Road was insolvent and needed coal cars before it could run. It contracted with the Proprietors of Locks and Canals on Merrimack River, whose machine shop in Lowell, Massachusetts, built the running gear for the cars. Historian Michael J. T. McMillen counts this 1842 Running Gear Contract among the first contracts in the lineage of equipment finance: it was the first of the railroad's 1842 car contracts to use a trust, a legal device that kept a claim on the equipment for the builder, and the first to set payment by the ton of coal hauled, and other car builders soon adopted the trust.
LessonEquipment finance began as a way to supply a buyer too weak to borrow, by keeping a claim on the machines themselves.
Credit
Philadelphia and Reading Rail Road (president Elihu Chauncey); Proprietors of Locks and Canals (treasurer Patrick Tracy Jackson)
1964
Leasing reaches Japan
In 1964 the trading house Nichimen and the United States Leasing Corporation set up Orient Leasing in Osaka, backed by Sanwa Bank, with ¥100 million of capital. Leasing was then very new in Japan, and the company grew slowly through the 1960s. It renamed itself ORIX in 1989 and became Japan's largest general leasing firm, by the account of the International Directory of Company Histories.
Credit
Nichimen; United States Leasing Corporation; Sanwa Bank
1973
Airlines rent their jets
International Lease Finance Corporation (ILFC) of Los Angeles began operations in 1973, "as a pioneer in the aircraft leasing industry" in its own words. In an operating lease the airline rents an aircraft for part of its life and hands it back, and the lessor carries the risk of what the plane will then be worth. By the end of 2013 ILFC had about 1,000 owned or managed aircraft and called itself the world's largest independent aircraft lessor by number of owned aircraft, when its owner AIG agreed to sell it to AerCap for consideration then valued at about $5.4 billion.
Credit
ILFC; AIG; AerCap
1981
Leasing enters China
Orient Leasing brought leasing to China in 1981, founding China Orient Leasing Company with China International Trust and Investment Corporation (CITIC) and Beijing Machinery and Equipment Corporation. State-owned enterprises leased machinery for old factories, and in 1984 China Orient Leasing wrote $40 million of contracts, three times its total of two years earlier, by the International Directory of Company Histories' account.
Credit
Orient Leasing (now ORIX); CITIC; Beijing Machinery and Equipment Corporation
1996
Same leases sold twice
On 28 March 1996 the US Securities and Exchange Commission (SEC) charged Bennett Funding Group, a finance company in Syracuse, New York, and its chief financial officer, Patrick Bennett, with selling more than $570 million of securities since 1991 in what it called a "Ponzi" scheme. The securities were meant to be assignments of office-equipment leases, that is, the right to collect a lease's payments. The SEC alleged that some leases did not exist, among them over $55 million of made-up New York City Transit Authority leases, and that some real leases were sold to investors who had already bought them.
LessonA funder buying lease payments needs a way to confirm that the lease exists and that nobody has bought it already.
Credit
SEC
2015
GE Capital sells up
On 10 April 2015 General Electric said it would sell most of GE Capital, its finance arm, within 24 months, including most of its commercial lending and leasing, and would work with regulators to end GE Capital's designation as a systemically important financial institution, a label that brings closer federal supervision. It kept three finance units tied to its own products, among them GE Capital Aviation Services (GECAS), the aircraft lessor. AerCap, a lessor based in Dublin, completed its purchase of GECAS on 1 November 2021 and said the combined company had over 2,000 aircraft, over 900 engines and over 300 helicopters.
Credit
General Electric (Jeff Immelt, Keith Sherin); AerCap (Aengus Kelly)
Watch for
Chips leased like jets
Bloomberg reported in October 2025 that a special purpose vehicle (SPV), a company set up for one deal, would raise about $20 billion of equity and debt, buy Nvidia graphics processors (GPUs) and lease them to Elon Musk's xAI for its Colossus 2 data center in Memphis, with Nvidia putting up to $2 billion into the equity. The debt would be secured on the chips.
Achieved whena party to the deal, a filing or a rating agency confirms that the vehicle owns the chips and leases them to xAI, and states the lease term.
Claimed by
Bloomberg, citing people familiar with the plan, October 2025
Open
01
Usage-billed leases, tested
A public test of the billing behind usage-based leases, in which a machine's payment rises and falls with the hours or miles its telematics unit reports, usually as a minimum payment plus a charge per hour or mile above an allowance. The test holds made-up monthly hour-meter feeds with the faults such feeds really have (a meter reset after an engine-controller swap, missing days, duplicate readings, a unit that reports minutes as hours, a machine idling all night), three common lease clauses (minimum payment with an hourly allowance, pay per hour with a cap, seasonal skip months), and for each case the invoice a careful person would issue.
The briefCredit · unclaimed02
Lease quotes worked by agents
A public test of whether AI agents can price an equipment lease the way lessors' quoting staff do: find the monthly payment that earns a target yield, or the yield that a payment earns, for leases with payments in advance, skipped or seasonal months, step payments, a security deposit, a residual value or balloon at the end, and stated tax assumptions. Each case has an answer computed by two independent programs.
The briefCredit · unclaimed
Open 01 · the brief
Usage-billed leases, tested
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public test of the billing behind usage-based leases, in which a machine's payment rises and falls with the hours or miles its telematics unit reports, usually as a minimum payment plus a charge per hour or mile above an allowance. The test holds made-up monthly hour-meter feeds with the faults such feeds really have (a meter reset after an engine-controller swap, missing days, duplicate readings, a unit that reports minutes as hours, a machine idling all night), three common lease clauses (minimum payment with an hourly allowance, pay per hour with a cap, seasonal skip months), and for each case the invoice a careful person would issue.
Why it is still open
Lessors already bill this way. Daimler Truck Financial's Dynamic Lease, announced in October 2019, bills a minimum payment plus a variable part from miles recorded by the truck's built-in telematics, and makers' finance arms sell hour-based leases for construction machines. Each bills from its own data with its own rules. This research found no public test of whether such bills come out right when the data go wrong; a GitHub search found only sample code for reading the ISO 15143-3 telematics standard, from John Deere and the German firm GEDA.
Built with what exists
The ISO 15143-3 (AEMP 2.0) data format, the industry standard for machine hours and locations, with GEDA's open documentation; Python and pytest; billing code written by hand and by AI agents such as Claude Fable 5.2 (see the Note "Three working devices") for comparison.
First step
Write 20 cases covering four faults, work out their invoices by hand, and ask one equipment-finance billing clerk to check them.
Done looks like
At least 200 feed-months covering at least eight kinds of fault and three clause types, each invoice worked out independently by two people who agree to the cent, a public scoring script, and results for at least two billing programs showing the share of invoices right to the cent and the money over- or under-billed.
Open 02 · the brief
Lease quotes worked by agents
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public test of whether AI agents can price an equipment lease the way lessors' quoting staff do: find the monthly payment that earns a target yield, or the yield that a payment earns, for leases with payments in advance, skipped or seasonal months, step payments, a security deposit, a residual value or balloon at the end, and stated tax assumptions. Each case has an answer computed by two independent programs.
Why it is still open
Lessors price with dedicated software; Ivory Consulting sells SuperTRUMP to "model and price complex, after-tax equipment leases and loans". Public tests of AI on finance ask other questions: FinanceBench, from Patronus AI (2023), holds 10,231 questions about public companies' filings, not about structuring a lease. This research found no public test of lease pricing by AI.
Built with what exists
Python with numpy-financial, and a LibreOffice Calc workbook as the second, independent answer key; AI models such as Claude Fable 5.2 (see the Note "Three working devices") and the open-weight Mistral Large 4 (see the Note "A trillion open parameters") as the agents under test, with and without a code tool.
First step
Write 30 cases, build both answer keys, and resolve every disagreement between them.
Done looks like
At least 300 cases across at least eight lease structures, the two answer keys agreeing on every case to within $0.01 of payment and 0.01 percentage points of yield, and a public table for at least three agents giving the share of cases each gets within those limits.