He doesn't solve the world's hardest decisions. He builds the language that lets a computer do it - across power grids, portfolios, and supply chains nobody can hold in their head.
When a utility decides how to dispatch power across a grid at 3 a.m., or a bank rebalances a portfolio against a thousand constraints, there is a good chance the math underneath was written in AMPL. Bill Wells is the CEO of the company that makes it. He is not a household name, and that is rather the point - optimization is invisible right up until the moment it isn't.
AMPL stands for "A Mathematical Programming Language." It was designed in 1985 at Bell Laboratories, the same lab that gave the world Unix and the C programming language. Wells now runs the small California company that turned that research artifact into commercial infrastructure used by enterprises, governments, and universities across more than forty industries.
His training fits the job. Wells holds a Master's in Mathematics from the University of California, Berkeley, earned in the late 1980s. Before AMPL, his path ran through the consulting firm Quovera and even a stint connected to Mountain View High School - a reminder that the people who end up running deeply technical companies don't always arrive by the obvious road.
The trick AMPL pulls off is translation. A planner thinks in messy, real-world rules: don't exceed capacity, meet demand, keep costs down, obey the regulator. A solver wants cold linear algebra. AMPL sits in between, letting you write models in something close to ordinary algebra and then handing them to whichever solver you like.
Write a problem in familiar algebraic notation instead of wrestling it into a solver's native format. The model reads like the math, not like plumbing.
AMPL is deliberately solver-agnostic. Commercial, open-source, linear, nonlinear, global - it talks to all the major engines and plays favorites with none.
APIs for Python, R, C++, C#, Java and MATLAB carry models out of the notebook and into live business workflows where decisions actually get made.
AMPL's customers span energy and utilities, finance, supply chain, transportation and tech. The bars below sketch where large-scale optimization tends to do the heaviest lifting - illustrative of the platform's reach, not a financial breakdown.
Bell Labs gave the world C, Unix, and AMPL. Wells is the steward of the third.
Employees running a platform used by enterprises, governments, and universities worldwide.
Co-founders - Fourer, Gay, and Kernighan - the third a co-creator of C and Unix.
Programming languages the AMPL API speaks: Python, R, C++, C#, Java, MATLAB.
The year a 1985 Bell Labs language formally became an independent company.
There's a nice symmetry to the company Wells runs. AMPL shares a birthplace with two of computing's most important inventions, and one of its three creators - Brian Kernighan - literally helped write the book on C. That heritage isn't a marketing line; it sits on the board of directors.
The throughline of Wells-era AMPL is a shift in posture: from a beloved academic modeling language into modern, API-driven infrastructure that businesses run every day. The pitch is decision intelligence - take the constraints a business already lives with and turn them into models that produce answers fast enough to act on.
That ambition explains the company's presence at decision-intelligence summits and its steady investment in connectors for the Python ecosystem. The goal isn't to win a benchmark. It's to be the layer optimization quietly runs on.