Leadership1 distinct publisher3 min readPublished
The employment-based backlog has passed 1.2 million people against a statutory ceiling of 140,000 a year, which means the preference category an employer files in now matters more to a skilled hire's tenure than the offer letter.
The Board Room · Leadership desk
Compiled by The Board RoomSomething wrong?How this is made
The ceiling is doing most of the work here. Federal law permits 140,000 employment-based green cards a year, and that number includes the spouses and children of the sponsored worker rather than sitting on top of them [10]. Divide the reported backlog by the annual allocation and the queue takes roughly 8.6 years to clear on the generous assumption that every card goes to someone already waiting and no new petition is ever filed [1]. Per-country limits then push that average around unevenly [2], which is how one statute yields a five-year estimate in one category and a date past the year 2200 in another [4].
The queue grows by accretion, and the FY 2016 figures show the mechanism. USCIS approved 47,601 EB-2 applications for Indian nationals that year, which the National Foundation for American Policy puts at 98,594 people once dependents are counted, while 4,407 people from India actually received permanent residence in EB-2 [4]. The difference, 94,187 people, joined the backlog in a single year, and NFAP reports similar increases in 2017 and 2018 from FOIA data covering FY 2016 through FY 2018 [2][5]. The binding constraint on a backlog built that way is the ceiling and the per-country limits, not the speed of adjudication [11].
The Philippine EB-3 line shows the same arithmetic in a form small enough to check by hand. NFAP estimates that backlog at over 39,000 as of December 2025 [8], and Filipinos received an average of 6,451 EB-3 green cards a year between FY 2016 and FY 2018 [9]. One divided by the other is about six years [3], which is what NFAP's own estimate for 2026 filings says [7]. A hospital recruiting nurses is working against a divisor rather than a service-level target [9].
The 179-year figure is a modeling exercise: NFAP frames it as a potential wait under current law [3], not a prediction that anyone will hold a place in line for two centuries. Even so, the number points to the right cause. Forbes reports that the source of the multi-year waits is the statutory arithmetic of the ceiling and the backlog, not the pace of processing at USCIS and the Labor Department [11], so administrative repair is not the lever that moves it. What the record here does not supply is attrition. We do not know how many of the 1.2 million people in the queue [1] leave the country, or when, and nothing in this analysis measures it.
This quarter's decision and this decade's exposure are separate questions. This quarter, the live choices are which preference category to file in and whether to begin labor certification now, given that the process can run two to three years before permanent sponsorship properly starts, usually while the employee holds H-1B status [11]. This decade, the constraint is supply: international students make up roughly 75% to 80% of full-time graduate students in AI-related fields such as computer and information sciences [12], so a team drawn from that pool inherits the queue regardless of how the company classifies immigration internally. Forbes reports the backlog already makes attracting and retaining talent harder and causes hardship for families [13]; the planning consequence is that retention for these hires is a decade-scale commitment of legal and relocation capacity, fixed at the moment of filing rather than at the next compensation review.
Ranked by verification strength, evidence, and original report placement.
A National Foundation for American Policy report states that a high-skilled foreign national from India with a labor certification application or employment-based immigrant petition filed in January 2026 or later has a potential wait time of 179 years in EB-2, 38 years in EB-3 and 5 years in EB-1.
Under U.S. law only 140,000 employment-based green cards are permitted annually, including the dependents (spouse and children) of the principal.
The employment-based immigration backlog in the United States now exceeds 1.2 million people, according to an analysis of U.S. Citizenship and Immigration Services data.
Wait times are much longer for individuals from India, China and the Philippines because of a per-country limit.
In FY 2016 USCIS approved 47,601 EB-2 applications for Indians, an estimated 98,594 people once dependents are included per NFAP calculations, but only 4,407 people from India were granted permanent residence in EB-2 that year.
NFAP reports similar increases to the backlog took place in 2017 and 2018, and obtained FOIA data for FY 2016-18 on permanent residence obtained by nationality and immigration category.
Distinct publishers with included, body-backed reporting in this cluster.
forbes.com
1 article · September 1, 2026
Follow any of these and your For You feed starts watching them — no settings page required.
invest
The 81% Problem: AI's Star CEOs Are Polling Badly With The People They Need To Hire1 distinct publisher
build
DHS put a $103,265 price on the student-to-H-1B hire, and its own math names who cannot pay1 distinct publisher
invest
Korea drops Calculus II from its 2028 exam and hands the gap to engineering faculties1 distinct publisher
invest
A whistleblower dates the USPS ballot portal build to 58 court-permitted days1 distinct publisher
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
One report, relayed once
The government data underneath is real — FOIA-obtained issuance by nationality and category, plus the August 2026 Visa Bulletin — and the numbers hang together: 39,000 Philippine EB-3 cases against 6,451 cards a year does come out at the six years Forbes reports. What holds the score down is that a single publisher is relaying a single organization's analysis, the total backlog and the Philippine count both depend on dependent ratios the report estimates rather than observes, and the enrollment figure for AI graduate programs is credited only to what recruiters 'find'.
Employer side unmeasured
Nothing in this reporting counts behaviour. There are no figures on petitions employers filed this year, no company disclosures about sponsorship policy, no attrition data, no record of a single hire lost to a queue. The story establishes the size of the line; it never observes anyone stepping out of it.
Math sound, conclusion ahead of it
A 179-year wait sounds like hyperbole and is not: it is today's issuance rate extended forward, and Forbes says as much by tying it to current backlog and bulletin data. The overshoot sits one step later. Calling this a retention problem is a reasonable inference from an 8.6-year arithmetic floor, but it is presented as established when no worker, employer, or departure appears in the evidence. Add the projection's own fragility — it assumes Congress never acts, which the piece elsewhere treats as the variable that matters.
One interested source, no counterweight
Every figure here originates with the National Foundation for American Policy, an organization publishing to make an argument about the cap, and Forbes closes on a senator blocking a green-card exemption for science Ph.D.s — a frame in which Congressional inaction is the cause and the remedy is obvious. That does not make the arithmetic wrong; the FY 2016 numbers are what they are. It does mean the choice of what to count, which dependent ratio to assume, and which years to average was made entirely by one party, with no agency comment and no analyst arguing the other side.
Firm on the ceiling, thin past it
We would defend two things without hesitation: the 140,000-a-year statutory ceiling counting dependents, and the fact that faster paperwork cannot fix a numerical shortage. Confidence falls as the claims travel outward — the modeled dependent counts, the century-scale extrapolations, and finally the unsourced enrollment share and the unmeasured retention effect. Single-publisher, single-report coverage caps how high this can reasonably go.