The air above Arlington looks comfortably vacant. No rush hour. No queue. No small argument over the last available seat. Yet every phone call, warehouse camera and private 5G radio enters an invisible city with zoning rules, protected tenants and a finite amount of useful room. Iyad Tarazi has spent more than a decade making that city less wasteful. His basic proposition is almost domestic: radios ought to be better housemates.
Tarazi is the president, CEO and co-founder of Federated Wireless, a company built around dynamic spectrum sharing. The phrase is dry enough to dehydrate a room. The practical idea is lively. Instead of assigning a valuable slice of air to one user everywhere and forever, software can coordinate access by place, time and power, while protecting incumbents such as military radar. Tarazi has called it the “Airbnb of spectrum.” The rooms were already there. The work was persuading the owners, visitors and building inspectors that the keys could move safely.
That required more than a clever controller. It required a market, a regulatory framework and an ecosystem of equipment makers, carriers, cloud providers and government users willing to behave as parts of the same machine. Tarazi had spent his career inside precisely those complications. His gift was not arriving innocent. He knew why large networks disliked surprises because he had been responsible for preventing them.
A drawing table, then every job in the room
He was born in Libya, where his father worked as a controller for a European car dealership in North Africa. At twelve, his family returned to the Gaza Strip, living in a refugee-camp area for about five years. Engineering already had a face: his grandfather, one of the early engineers in Mandate Palestine. Tarazi remembers sitting beside his drawing table, looking through designs and feeling proud of the profession before he had entered it.
At sixteen, Tarazi and his brother flew to the United States with $500. He earned a bachelor's degree in electrical engineering from the University of Maryland and later a master's in engineering management from Southern Methodist University. The neat credentials conceal the sharper lesson he drew from those years: people regularly underestimate what they contain and what they can do.
“I think a lot of people underestimate what is inside them and what they can do.”Iyad Tarazi
His first job after college was at a software startup of roughly twenty people competing in the young relational-database market. His title could not protect him from the entire company. He managed C++ code, answered technical-support questions, handled marketing, packed trade-show boxes and carried them to the show. It was an excellent cure for believing that product and business are neighboring countries.
The scale changed dramatically. Tarazi held positions at MCI, then ran network engineering at Nextel, overseeing planning, integration, performance engineering, testing and core deployment. After the Sprint-Nextel merger, he led large R&D and modernization programs at Sprint. His responsibilities reached across cloud software, spectrum, radios, transport, applications and devices. He organized teams numbering in the hundreds and managed investments measured in hundreds of millions of dollars.
The small company with the inconvenient idea
By 2014, the size of Sprint had become part of the problem. Tarazi wanted to build again. He had prepared with unusual deliberateness, serving for several years on the board of a friend's startup to learn the questions that mattered before he took responsibility for his own. Then came Federated Wireless, an early-stage company formed around spectrum-sharing research by Virginia Tech scientists. The scientists had significant intellectual property and little interest in becoming operators. Tarazi found the arrangement intensely appealing.
The problem had bothered him for years. Indoor wireless coverage was distributed across too many buildings for a carrier to solve neatly. More broadly, spectrum was allocated in large exclusive blocks even when much of it sat unused in particular places or times. A cloud-based coordination layer could grant access quickly, protect existing users and let enterprises take more control of their own connectivity. The software was only one part. Someone had to turn a new technical model into something cautious organizations could buy.
Early conversations supplied a brutal education. In one meeting, industry participants shouted at the Federated team, called the proposal foolish and asked who they thought they were. Tarazi laughed, went for a run and reconsidered. A meaningless idea would not have produced such heat. The reaction revealed both danger and value. Federated refined its approach and worked patiently across the ecosystem. Eighteen months later, the company returned to that office with a partnership agreement.
To simplify, you have to know too much.Tarazi on why varied experience matters
That sentence captures his operating method. Shared spectrum could not be sold as an elegant rebellion against the old system. Federated needed to understand the old system well enough to preserve what mattered: reliability, incumbent protection, regulatory accountability and equipment interoperability. Disruption, in Tarazi's hands, came with checklists.
Software startup work spanning code, support, marketing and trade-show logistics.
Network engineering, large-scale R&D and modernization across the wireless stack.
Joins Federated Wireless to commercialize dynamic spectrum sharing.
Commercial CBRS deployments begin at scale, turning years of coordination into operating infrastructure.
Spectrum AI enters general availability, joining radio physics with continuous network data.
The culture mirrors the product
Tarazi says his true passion is not simply creating products and services. It is creating an environment where he wants to work. At Federated Wireless, he has described broad access to company information, weekly town halls where decisions are shared and little appetite for status built from informational privilege. He once said he had not looked at the org chart since the company started.
There is an agreeable symmetry here. A company asking radios, regulators and rivals to coordinate cannot sensibly worship private fiefdoms inside its own walls. The culture he describes prizes diverse technical and commercial knowledge because the problem refuses to stay in one department. In 2019, employee ratings from people of color placed Federated Wireless among Comparably's small and mid-sized companies recognized for CEOs supporting diversity. In 2020, Tarazi received Mobile Breakthrough's CEO of the Year award. He redirected the latter recognition toward the team's work.
A better map of the invisible
The next chapter returns Tarazi to a familiar nuisance: uncertainty. Traditional radio-frequency planning simplifies the physical world because tracing every interaction among signals, buildings, terrain, water and foliage has been computationally expensive. When engineers cannot predict precisely, they protect the network with conservative margins. They reserve capacity, reduce power, build extra sites or spend time testing in the field. Every hedge is reasonable. Together, they are expensive.
Federated Wireless now describes Spectrum AI as physical AI for shared-spectrum planning and coordination. It combines GPU-accelerated ray tracing, three-dimensional geospatial information, equipment data, live spectrum assignments and field measurements. The aim is to simulate how radio waves move through actual places, then keep learning as networks and environments change.
Tarazi remembers when a comparable network plan required two weeks and a server farm. In 2026, he described running the work in seconds. Field deployments reported by the company showed gains of up to five times in network capacity, up to 50 percent more usable spectrum capacity and propagation predictions above 90 percent accuracy. These are “up to” figures, not promises to every operator. Their strategic meaning is still clear: better prediction can change the capital plan before money hardens into towers.
It also widens the questions network engineers can afford to ask. Which rooftop is better? What changes if an antenna moves one meter? How many homes can a proposed site actually serve? Where will interference turn a business case sour? A faster model is useful. A model fast enough to compare many possible businesses is something else.
“We'll never be done. You have to constantly update your data and constantly run your models.”Iyad Tarazi on the learning system
“Never done” is an uncommonly honest slogan for AI. Trees grow. Equipment changes. A neighbor lights a new radio. The model encounters the same unruly physical world that humbled every earlier planning tool. Tarazi's aspiration is not to abolish uncertainty with a theatrical flourish. It is to reduce it enough that operators can use more of what they already own.
The fourth lever
Wireless operators have traditionally grown capacity in three expensive ways: buy spectrum, build infrastructure or upgrade equipment. Tarazi argues for a fourth lever, extracting more performance from existing assets through better physical intelligence. It is the same instinct that brought him to Federated Wireless. Apparent scarcity may hide a coordination problem. Apparent capacity limits may hide a prediction problem.
His journey gives that thesis unusual weight. The child beside a drawing table learned to respect engineering. The twenty-person startup taught him to pack the boxes. Nextel and Sprint taught him how large networks resist risk. Federated Wireless taught him that an ecosystem has to be built alongside a product. The work now is to make invisible infrastructure legible enough that people can place better bets on it.
The air will remain crowded. Tarazi seems content with that. Crowds are not failures when someone has designed the room.