AudioCodes

The founding problem was arithmetic. A telephone call digitised the standard way consumes 64 kbps, and a packet network of the early 1990s could not carry many of those. Everything that became voice over IP depended on compressing speech hard enough to fit, without the result sounding like a machine. That is a signal-processing problem before it is a networking problem, which is exactly why the founders were chip people rather than network people.

And their codec work went into an international standard. AudioCodes' speech coding contributed to G.723.1, adopted by the ITU - the low-bit-rate coder that a great deal of early VoIP ran on. A company's own specification becoming the standard everybody implements is the second instance of that pattern in as many entries on this timeline: Anixter's Levels programme became the TIA's Categories, and this became an ITU recommendation.

Which is worth naming as a mechanism rather than treating as two coincidences. Standards bodies rarely invent; they ratify. What they ratify is usually the thing already built by whoever needed it badly enough to build it, and the reward for that company is not the licence income - it is that the industry's default now has your fingerprints on it, and every competitor implements a thing you understand better than they do. The moat is not the patent. It is the head start.

The product line climbed one layer at a time and the sequence is unusually clean: codec, then the board carrying the codec, then the media gateway carrying the boards, then the session border controller carrying the gateway's function in software, then the management and analytics around it. Most component companies get commoditised as their layer becomes a feature of the layer above. This one kept moving up before that happened, which is a strategy easy to describe and hard to execute, because it means competing with your own customers.

Then a structural fact worth noting. Adlersberg has been chief executive since 1993 - a thirty-three-year run, among the longest of any founder still leading a listed company. Its peers in enterprise voice cycled through leadership repeatedly and, as one analysis puts it, oscillated between growth mandates and restructuring mandates. A company that changes chief executive every four years cannot run a strategy that takes ten, and the enterprise-voice graveyard is full of businesses that were three years into a transition when the person who started it left.

The Israeli lineage is the other half of the context. Adlersberg co-founded DSP Group in 1987, built a digital cellular research team there that was spun out as DSP Communications in 1992, and founded AudioCodes the following year. Three companies from one engineering group inside a decade - which is what a functioning technical ecosystem looks like from the inside, and is the same shape the Kentik entry describes for network operations in California.