Amir Husain and the computers that refuse to retire

The first computer Amir Husain built could not calculate anything. Its components were broken toys and cardboard boxes. In Lahore, after seeing a Commodore 64 run Hangman at a friend’s house, he went home and assembled his own approximation. The machine was imaginary. The fascination would prove considerably more durable.

He remembers being four. What caught him was the creative possibility of a box that could do something unexpected. Hangman is a modest demonstration of computing, unless you have never encountered a computer before. Then a machine that appears to know a word can rearrange the furniture in your head.

Years later, that child became the founder of SparkCognition, the industrial AI company now called Avathon. He became founding CEO of SkyGrid, a venture with Boeing. He wrote books about artificial intelligence and helped fund university research. Somewhere along the way, he also accumulated the sort of computer collection that makes a person hesitate before describing a workstation as obsolete.

The collection provides a useful way into his story. Husain’s work repeatedly returns to the unused possibilities inside a system: a machine, an algorithm, a research community. He has changed roles and companies. The question has survived.

Machines with expensive responsibilities

Husain earned his computer science bachelor’s degree at the University of Texas at Austin in 1998. In 2013, he founded SparkCognition. Its territory included industrial operations, where a useful prediction might concern machinery rather than the next sentence in a chat window. These are environments in which an error can acquire a repair bill very quickly.

His examples make the work tangible. Could a computer detect the approach of a failure in a $100 million turbine? Could it help anticipate trouble on an offshore oil rig? The appeal lay in finding patterns difficult for people to see early enough to act. A prediction has a practical deadline: before the equipment fails.

Industrial AI also demands an acquaintance with the world beyond software. Sensors produce imperfect observations. Equipment has a history. A warning must mean something to the people responsible for a plant or a fleet. The interesting work begins when an algorithm meets those obligations. In Husain’s career, computing repeatedly leaves the screen and takes a job among physical things.

That financing arrived while Husain was leading the company. The announcement put total capital raised at $300 million and described investment in research, development and expansion. It also pointed to HyperWerx, a 50-acre facility for demonstrating and testing technologies including robotics and autonomous drones. A business built around software had acquired room to move.

Husain’s current biography dates his CEO tenure through the end of 2023. SparkCognition adopted the Avathon name in 2024. Today, Avathon lists Pervinder Johar as CEO and Husain on its board. The distinction matters in a career with so many overlapping affiliations: founding a company and running its daily operations are different chapters.

The sky had a software problem

In November 2018, Boeing and SparkCognition announced SkyGrid, with Husain to serve as CEO. The premise involved autonomous cargo and passenger aircraft sharing airspace. The aircraft attracted the imagination. The less visible challenge was arranging for all those machines to get along.

The venture proposed software combining AI, dynamic routing, analytics, security and blockchain technology. Its announced uses included package delivery, industrial inspection and emergency assistance. Each depended on more than making a vehicle fly. A useful flight needs a route, a context and a way of responding when conditions change.

This is an unusually revealing episode in Husain’s working life. His attention extends from the clever machine to the arrangements that let it operate. The skies offer little sympathy for a product demonstration that has neglected the surrounding system.

Autonomy also brought him into questions of defense and public policy. He founded SparkCognition Government Systems, whose chair was former U.S. Deputy Secretary of Defense Robert Work. His advisory work has included NATO’s maritime unmanned systems board. In public discussion, he has argued that military AI demands serious attention because other countries will pursue it, while also raising questions about ethics and explainability. Those are his arguments about the technology’s direction, with responsibilities attached.

A laboratory outlasts a product launch

There is another way to follow a founder’s priorities: look at what he helps other people build. In 2020, Husain and his wife, Zaib, gave $5 million to launch UT Austin’s Machine Learning Laboratory. Both are alumni. She graduated in business administration in 1999; he completed his computer science degree a year earlier.

The laboratory provided a permanent base for AI research across the university, including the NSF AI Institute for Foundations of Machine Learning. The gift supported students, faculty and work centered there. It was an investment in a community whose future questions could not all be specified in advance.

Husain has explained his continuing attachment to UT in practical terms. Computer science supplies enabling technology, while other disciplines contribute knowledge of the problems. Austin gave him access to that combination at scale. The relationship places the university inside his entrepreneurial story, as an active source of collaborators and ideas.

In May 2025, he returned to address the College of Natural Sciences graduating class. The former student was now the commencement speaker. Universities are quite good at arranging these pleasing reversals. The more consequential return had already happened through the laboratory.

Please do not throw the workstation away

Old computers are a particularly personal part of this relationship with technology. Husain traces the SGI machines in his collection to using SGI O2 workstations in UT’s computer science labs in 1998. Their Unix environment differed from the DOS computer he had used at home. A working environment became a memory worth preserving.

His ATX Computer Museum project presents machines together with software, photographs and records of their use. Its collection includes an SGI Indigo presented to Steven Spielberg during the making of Jurassic Park, and a SOL-20 from Paul Allen’s collection. The project lists a public opening planned for February 2027 in Florence, Texas.

A computer on a shelf can resemble an unusually heavy ornament. Software returns its purpose. The museum’s attention to operating environments and surviving records makes room for the people who used these systems and the particular things they were trying to accomplish. The machines have biographies, too.

“What can this do that nobody thought it could do?”

Amir Husain, on his old computers

The question also fits Kindi, a software project on his website. It describes running modern language models on NVIDIA graphics cards ranging from an older Quadro K420 to an RTX 3080. Its published measurements include tests on a card with just one gigabyte of memory. These are the project’s own results, tied to particular models and settings.

There is a pleasing continuity here. The collector keeps an old machine available for inspection; the programmer looks for work it can still perform. A specification can describe a limit without exhausting the possibilities. Husain seems inclined to investigate the remaining space.

The author keeps opening the casing

His writing moves between the details of computing and the institutions built around it. The Sentient Machine introduced a general readership to his thinking about AI. Hyperwar, which he co-authored, examined conflict and competition. His catalog also includes books on generative art, leadership and computing history.

Published by Basic Books in August 2025, The Cybernetic Society develops an argument about humans and intelligent machines becoming intertwined. Husain presents that relationship as a source of opportunity and danger, with consequences for how societies make decisions. It is a claim about institutions as much as about devices. A machine’s output matters because someone, or some system, acts on it.

His 2026 title Algorithms from Scratch turns toward the machinery itself. It works through computing ideas in self-contained Python programs, including search, optimization and simulation. The educational premise is straightforward: understand what the code is doing. Alongside his broader arguments about AI, it gives readers a route into the mechanisms.

He has also spoken about enjoying philosophy and poetry. That interest helps explain the breadth of the questions around his technical work, without supplying a tidy answer to them. How a system functions and what people ought to do with it remain separate subjects requiring attention.

Another question, another machine

Husain now chairs WorldQuant Foundry, a venture studio supporting new companies. The work extends his company-building experience into helping other founders. His public activities also include software projects, books and talks. In July 2026, his channel published his convocation address at the Lahore University of Management Sciences, bringing the geographical story back toward its beginning.

Asked about the ingredients of his career, he has credited encouraging parents and warned against treating effort as a guarantee: “Hard work is a necessary, but not a sufficient condition.” It is a useful reservation in a founder’s story. Curiosity needs opportunities, collaborators and circumstances in which it can become work.

The cardboard computer belongs near the start because it captures something that survives all the later machinery. Husain saw a possibility before he could reproduce it. Decades later, the equipment is real, the questions are larger, and he is still looking for another thing to make it do.