The State of US Oil & Gas Production Operations
America runs on a vast, aging, low-rate well base. Almost a million producing wells, four in five of them marginal, decline curves that punish every day of delay, and an operating cost base in the tens of billions. The opportunity is not more drilling. It is running what is already in the ground better. Every figure below is cited to its primary source.
Almost a million wells, down from the 2014 peak
The United States had 930,445 producing oil and gas wells in 2023, down from a peak of 1,031,161 in 2014. The count is falling, but the absolute number is staggering: roughly a million distinct revenue-and-risk assets that someone has to watch, visit, repair, and report on, every single day.
This is the operating problem in one number. Whatever you want to do better, you have to do it across the better part of a million wells, with crews and budgets that are not growing.
Four in five wells are marginal, and they barely register in output
In 2023, roughly 78% of US wells produced 15 barrels of oil equivalent per day or less, a share that held near 80% for two decades. By count, the field is dominated by marginal and stripper wells; by volume, that long tail contributes a small fraction of national output.
That gap is the whole game. A low-rate well dies the moment its operating cost crosses its revenue. Whether it lives another decade or gets plugged this year is decided not by geology but by how efficiently it is operated. This is where lease-operating-expense discipline turns marginal wells back into contributing ones.
Every well is born dying
In the most prolific tight-oil plays, wells typically decline more than 50% in their first year and roughly another 30% in the second, with most of a well's lifetime production landing in its first two years. Output races downhill from day one.
Picture the decline curve next to a flat cost line. As rate falls, cost per barrel climbs. The well dies the moment cost crosses price. Operations discipline (the right work, on the right well, on the right day) pushes that crossing further out. It is the difference between plugging a well and keeping it producing. See how that ranking works in pump by priority and decline-curve analysis.
Upstream operating cost runs to the tens of billions a year
No single primary source publishes a clean national upstream lease-operating-expense (LOE) total, so we build one transparently. Start from a hard, cited input (US crude production) and multiply by a conservative, industry-typical LOE-per-barrel range. The math is below; the result is a labeled WorkSync estimate, not a third-party figure.
WorkSync estimate. Inputs: EIA-reported 2024 US crude production, annualized, multiplied by a stated LOE-per-barrel range. LOE per barrel varies widely by basin, well type, and lift method, and this figure excludes natural-gas operating cost, so treat it as an order-of-magnitude floor, not a precise total. The point stands either way: the prize is in the operating line, and a few percent of it is real money.
An aging network, with constrained new construction
The US natural gas pipeline network spans about 3 million miles of mainline and other pipelines connecting production, storage, and consumers, including roughly 300,000 miles of higher-pressure transmission line. Much of it is decades old, and new construction stays constrained by permitting and capital discipline.
That means the operating burden is not just wells. It is the gathering and transmission network feeding them, where integrity, modeling, and inspection work compounds the same efficiency problem. WorkSync references 4,000+ miles of pipeline in deployment, and FlowSync handles the hydraulic-model side of that work.
AI has transformed software. It has barely touched the oilfield.
The Anthropic Economic Index (March 2026) maps where AI is actually being used. Adoption is heavily concentrated in software: Computer & Mathematical tasks make up about 35% of Claude.ai conversations. Even there, Claude still covers only about 33% of the tasks in that single highest-adoption category, so the ceiling is far above current usage.
Physical and production occupations sit at the bottom of the curve, far behind knowledge work. The engineering and field-operations work that runs oil and gas is exactly the kind of high-value task where the gap between what AI can do and what is actually being done is widest. That gap is the opportunity: operators who close it early compound an advantage that widens every quarter. See the operator framing in AI for upstream.
Independents run the field, and the efficiency math scales there
Approximately 9,000 independent producers operate the large majority of US wells and account for about 83% of US oil production and roughly 90% of natural gas production. The supermajors are the headline; the independents are the base.
That matters because independents run lean. They do not have armies of data engineers to build custom optimization platforms. They need the operating-efficiency edge as a product, ready to run on the systems they already own. That is the gap WorkSync was built to close.
The supermajors solved this by hand. WorkSync productized it.
A million wells, four in five of them marginal, on steep decline curves, against a tens-of-billions operating cost base, run mostly by lean independents, with the AI tools that transformed software barely deployed in the field. Every line in that sentence points the same direction: the prize is operating discipline, not more drilling.
The industry already named the idea. Pump by exception (visit only the wells that deviate) has been the goal for over a decade, and adoption has struggled because doing it well takes infrastructure most operators cannot build. The supermajors built it by hand, at great expense.
WorkSync productized it as pump by priority: WellOPS ranks every well by cash flow, routes the day around value density, and learns from every outcome, running on the SCADA, ERP, and CMMS you already own. The reference deployment spans 5,000+ wells across three basins, with 15% free cash flow uplift on the same crew, 35% fewer site visits, and TRIR cut from 1.8 to 0.3.
Explore WellOPS, pump by priority, and FlowSync.