The story of Cohere, as framed in a new Light Reading analysis, doubles as an indictment: the open RAN movement has not yet delivered on the promises that carried it through its loudest years.

That is a hard claim to sit with, because open RAN arrived with specific, measurable commitments — interoperable radios from multiple vendors, lower equipment costs, and a break from the tight coupling between baseband and radio hardware that defined the four-vendor market dominated by Ericsson, Huawei, Nokia and ZTE. Cohere's trajectory, Light Reading argues, illustrates where those commitments have quietly buckled.

The shorthand deserves unpacking. Open RAN refers to the effort — driven by the O-RAN Alliance's specifications and, in Western markets, by regulators and legislators wary of Huawei — to standardize the interfaces inside the radio access network so that operators can mix components from different suppliers. Its credibility rests on a simple test: does a small or specialist vendor actually win deployment at scale in a major operator's network, on commercial terms, without being absorbed into a single-vendor wrapper?

By that test, the record is thin.

What does the Cohere case actually show?

Cohere Technologies, a US-based software and waveform specialist, spent years promoting a mu-MIMO scheduler it argued could multiply the capacity of existing cellular spectrum without new radios. The pitch was seductive: more throughput from assets operators already own. Yet the company's path — from ambitious standalone ambition to partnerships with, and eventual deep entanglement with, the very large vendors open RAN was meant to bypass — tracks the broader pattern Light Reading identifies.

Small vendors rarely fail outright. They get partnered, funded, acquired or quietly shelved. The open RAN architecture that was supposed to create a level playing field has instead pushed specialists into alliances with incumbents, because integration costs, testing burdens and operator procurement habits still favor the big radio vendors. The result is an ecosystem that is open in name and concentrated in practice.

This is not a new complaint, but the Cohere story gives it a concrete face. The promise was displacement of the oligopoly; the delivery, so far, is a suite of niche suppliers orbiting it.

Where open RAN has fallen short

The shortcomings Light Reading points to are structural rather than incidental:

• Integration still costs more than the interfaces save. Standardized interfaces exist on paper, but operators bear the systems-integration burden when they mix vendors, and few have found the economics convincing outside flagship trials. -- Procurement habits die hard. Large operators, burned by integration risk, default to a small number of trusted end-to-end suppliers even in networks formally built on open principles.
• Small vendors lack the balance sheet for carrier-grade support. A scheduler or radio unit must be maintained across software releases, security patches and hardware generations — a burden that favors incumbents regardless of technical merit.
• Political tailwinds masked commercial weakness. Much of open RAN's momentum came from Huawei bans and Western industrial policy, not from operators independently concluding that open interfaces cut costs. Remove the policy premium, and the commercial case narrows.

None of this means open RAN is dead. The O-RAN Alliance's specifications continue to mature, and operators in Japan — where Rakuten built a network around the principle — and parts of Europe and India keep multi-vendor deployments in production. But those deployments coexist with a growing recognition that openness, by itself, is not a business model.

Is the criticism fair to the technology?

Partly. Open RAN's defenders can fairly note that the movement is young by telecom-equipment standards, and that virtualized RAN — a sibling trend — has made more tangible progress than the multi-vendor radio dream. They can also point out that Cohere's particular technology, a software scheduler, was always a harder sell than a commodity radio unit, and that one company's arc proves a pattern only tentatively.

Yet the skeptic's rebuttal writes itself: if roughly a decade of standardization, government pressure and operator enthusiasm has not produced a thriving population of independent RAN suppliers displacing the incumbents' share, the burden of proof has shifted. The question is no longer whether open RAN can produce trials and press releases. It is whether it can produce procurement contracts — and vendors that survive on their own balance sheets.

What comes next

Light Reading's read is a warning, not an obituary. The open RAN project continues, and its standards work on interfaces, RAN intelligence controllers and energy management feeds into mainstream vendor roadmaps even where the multi-vendor vision stalls. For vendors like Cohere, the realistic futures are partnership, acquisition or gradual irrelevance — the three exits that have absorbed most of the specialist field to date.

The movement's next test will be written in purchase orders, not position papers: whether operators, in their next round of RAN investment, actually award meaningful volume to independent suppliers — or whether openness remains, as Cohere's story suggests, a label attached to networks whose economics still run through the same handful of names.