There is a funny little time warp at the edge of a modern network. Inside one operator's domain, software can measure traffic in real time, calculate paths, apply policy, and move packets faster than any human can blink. Then the traffic reaches another operator. Suddenly, the future may require a spreadsheet. Abilash Menon has spent enough of his career around this border to recognize it as more than an inconvenience. It is the missing machinery between networks that work beautifully alone and an economy that needs them to work together.
Menon is a networking engineer, named inventor on more than 40 patents, and CEO and co-founder of MaiaEdge. The company sits in Burlington, Massachusetts, a short drive from his home base in Boxborough. Its premise is neatly expressed in one of his lines: “The internet connected applications and users, but it left networks negotiating by spreadsheet.” The sentence is both diagnosis and sales pitch. It also contains the quiet comedy of telecom, an industry capable of bending light through glass while asking two businesses to coordinate the handoff by email.
A career spent approaching the boundary
Long before MaiaEdge had a name, Menon's work kept circling the same question from different altitudes: how should a network decide where traffic goes? At Cisco, he led teams working on Multiprotocol Label Switching, the technology better known as MPLS. It gave large networks a disciplined way to steer traffic along labeled paths. If the public internet resembles a city where every driver reads street signs, MPLS is closer to a railway system, with routes decided and marked ahead of time.
He later joined 128 Technology, the Burlington networking startup that developed a session-aware approach to SD-WAN. There, the problem shifted. Enterprises were connecting branches, data centers, and services across mixed links, and conventional routing carried plenty of inherited ceremony. Menon's patent record from that period runs through source-based routing, performance monitoring, forwarding enforcement, and the mechanics of building more useful context into the journey of a packet. In 2019, he also gave a whiteboard series on the company's Services and Topology Exchange Protocol. The title sounds like an excellent cure for a crowded room. The subject, however, was practical: how network systems exchange enough knowledge to choose routes around services rather than merely addresses.
Juniper Networks acquired 128 Technology in 2020. Menon continued into large-scale network design and engineering leadership there. A pattern had formed across three chapters. MPLS made paths deliberate. SD-WAN made connectivity more adaptable. Session-aware routing made policy more closely reflect what an application needed. Yet every improvement still encountered the old boundary when a path crossed between owners.
While deepening the engineering case, Menon added the commercial vocabulary. He attended Northwestern University's Kellogg School of Management from 2021 to 2023 and completed its Executive MBA program. It was an apt interval for someone moving from designing how systems exchange information to asking why companies fail to exchange it.
“Networks always negotiate, they do not coordinate.”Abilash Menon, speaking at ITW 2026
Where elegant systems acquire awkward paperwork
Menon and co-founder Tim Ziemer started MaiaEdge in 2024. Their target is what Menon calls “the big middle,” the territory between data centers and enterprises where fiber operators, carriers, colocation facilities, cloud on-ramps, and regional service providers must cooperate. No single operator reaches every office, GPU cluster, partner, or cloud region. Extending a service therefore requires a chain of agreements and configurations across companies that use different tools and keep different policies.
The technical delay is only half the trouble. Once traffic leaves one operator's footprint, visibility can fade. The first operator still owns the customer relationship and the service promise, yet may lose a clear view of latency, jitter, packet loss, and the exact path. Federation must solve for trust as well as transport. Participants need to cooperate without handing a central platform their customers, margins, policies, or brand.
MaiaEdge's architecture places a Path Border Controller, or PBC, at a network boundary. A cloud-native Path Computation Engine, or PCE, calculates paths and supplies end-to-end visibility. The system is designed to travel over existing transport and merge Layer 2 and Layer 3 behavior at the edge. In plain language, operators can add a coordination layer beside infrastructure they already own, advertise capacity, activate private paths, and see how those paths perform.
The sovereignty idea is central. MaiaEdge does not sell the end customer a network service in competition with its operator customers. It gives operators infrastructure to build those services under their own names. The distinction sounds fussy until one notices the incentive. A carrier is more likely to federate if federation expands its reach without turning its customer into somebody else's account.
The machines are spreading; the seams remain
Artificial intelligence gives Menon's old networking problem a new clock. Training can be concentrated in a giant cluster. Inference, the moment a trained system responds to a request, follows users, data, available power, and cooling. That pushes compute into a patchwork of smaller locations. Some belong to hyperscalers, some to specialist AI operators, and some may be modular facilities beside an energy source. The resulting system crosses more facilities and more networks.
“AI doesn't care who owns the network,” Menon said at ITW. “It only cares I want to go from A to Z and get the data.” The line removes corporate boundaries from the packet's perspective, where they were never especially interesting. From the carrier's perspective, ownership still matters enormously. Every participating operator must preserve control while the workload demands an experience that feels continuous.
AI agents sharpen the requirement. A single task may trigger many calls among models, databases, tools, and services. The slowest hop can set the pace for the whole exchange. Bandwidth remains important, but predictability joins it: latency, jitter, route, and the ability to diagnose trouble across a path with several owners. Menon describes the network as the nervous system of the AI economy. It is a metaphor with an engineering bill attached.
“AI doesn't care who owns the network. It only cares I want to go from A to Z and get the data.”Abilash Menon at ITW 2026
The company announced a $20 million Series A in February 2026, led by executive management alongside G20 Ventures, with employees and individual investors participating. Menon avoided the usual ceremonial excitement in his own announcement. He wrote that the money would support the infrastructure and a measured expansion of the go-to-market team. His framing was deliberately long term: trends expire, while the need for independent networks to coordinate remains.
Standards rooms, hallway conversations, and beautiful boxes
A federated architecture cannot succeed as one company's private language. Menon has consequently taken the idea into Mplify, the industry alliance formerly known as MEF. He has participated in standards work, served as a session lead on agentic service orchestration, and urged operators to bring practical cases into the room. The work is slow in the useful way: governance, identity, deterministic interfaces, auditability, and the procedures that let competitors cooperate without pretending they have become friends.
His conference reports reveal a personality comfortable mixing precision with small jokes. After returning to Boston from Lisbon by way of Iceland and Seattle, he observed that a networking person appreciates path diversity, though this instance may have been excessive. At another event he called the MaiaEdge PBC possibly the most beautiful device ever built for the service-provider industry, then conceded that founders are generally advised not to review their own work.
That humor matters because the subject can disappear beneath acronyms. Menon's public interests also refuse to fit neatly inside a rack. He dances, follows chess, explores cuisines, and makes his own recipes. Cooking is an irresistible metaphor for federation, so let us resist it only briefly. Separate ingredients retain their character, timing governs the outcome, and someone must understand what can safely share a pan.
In June 2026, Menon joined leaders from carriers, cloud infrastructure, security, and data centers on a CCT Global panel in Dublin about network performance, security, and sovereign AI. Earlier that month in Lisbon, he had presented on distributed AI and joined two days of standards work. These are useful clues to his aspiration. MaiaEdge is selling equipment and orchestration software, but Menon is also trying to recruit an ecosystem into a shared operating model.
Look for the spreadsheet between two sophisticated systems
The founder lesson in Menon's story is pleasantly unglamorous. Important infrastructure opportunities often hide at organizational seams. Each side may be technically advanced. The interface between them survives through habit, paperwork, and a heroic employee who knows whom to call. It looks too mundane to deserve a company until a new workload turns the delay into a constraint.
Menon did not arrive at this seam through a fashionable detour. He followed routing from labeled paths to software-defined networks, from application context to inter-provider coordination. More than two decades of engineering gave him a close view of the recurring failure. Business school gave him another way to frame the incentives. A co-founder with long experience selling networking systems gave the technical thesis a route into the market.
Whether federated private networking becomes a standard layer will depend on adoption, interoperability, and the willingness of operators to make shared rules. The company still has to persuade a conservative industry that another box and control plane reduce complexity rather than add to it. Menon's answer is to make the equipment quick to deploy, the path visible, and the commercial boundary explicit.
There is something almost Wildean about the underlying problem: everyone agrees that networks ought to connect, provided no one loses control by doing so. Menon's architecture treats that contradiction as a design constraint. The packet gets a path. Each operator keeps its sovereignty. And, if all goes according to plan, the spreadsheet finally gets a quiet retirement.