The Wavelength Backlog That Wasn't: What Cogent's Lawsuit Reveals About the Real Business

Cogent Communications' optical wavelength backlog sounded like a growth engine. It turned out to be something much closer to a wish list.

The company told investors, repeatedly, that its backlog of booked-but-not-yet-provisioned wavelength orders was a forward-looking signal of strong demand. A backlog is supposed to be an order queue: customers have committed, Cogent provisions the service, revenue flows. Except in practice, these orders sat there for months — sometimes over a year — and then vanished. Customers who had signed up pushed out their acceptance dates, or just never showed. Cogent could build the pipe, but nobody moved in.

On May 4, 2026, CEO David Schaeffer conceded it on an earnings call: customers were "pushing out their acceptance of wavelengths" even though Cogent was provisioning more capacity than ever. The backlog, the company had assured shareholders, was an indicator of expected growth and stock value. It was actually an indicator of how much of the backlog would not convert.

A securities class action lawsuit, filed in July 2026, says Cogent knew or should have known that the vast majority of those orders were unlikely to become revenue. The complaint covers the period from February 29, 2024 to May 1, 2026 — a stretch during which CCOI stock fell from over $86 to under $17, a drop of more than 80%. The lead plaintiff deadline is September 21, 2026.

But the lawsuit is just the accounting of what went wrong. The more useful question is what the business looks like now that the story about wavelength growth has been stripped down to the actual numbers.

To understand why backlog is a fragile metric, you need to understand the product. Cogent's optical wavelength service is high-bandwidth, low-latency data transport — think of it as leasing a dedicated lane on a fiber-optic highway, connecting data centers across the U.S., Canada, and Mexico. Cloud providers, AI companies, content firms, and cable operators all need this stuff to move data between locations. The long-haul market is over $4 billion annually, dominated by Lumen and Zayo, which control roughly 90% of it.

Cogent entered this market in 2023 after acquiring Sprint's wireline fiber network from T-Mobile — about 19,000 route-miles and access to roughly 800 carrier-neutral data centers. The idea was elegant: Cogent already had the sales force, the IP network, and the fiber. Now it could sell wavelength contracts at prices like $2,000–$3,000 per month for 100 Gbps connections, typically on 1-to-3-year contracts.

The backlog was the bridge between "we have the network" and "we're making money from it." When Cogent reported a backlog of 3,400 orders in late 2024, it implied a massive pipeline of future revenue. The problem is that backlog is a commitment-to-serve, not a commitment-to-pay. The customer signs an order, Cogent provisions the physical connection at both data centers, and then the customer has to accept it and start using it. Between signing and accepting, there's a gap where the customer can delay, renegotiate, or walk away entirely.

The company said it could provision wavelengths in about two weeks — far faster than the 3-to-9-month provisioning cycle of competitors. But fast provisioning doesn't matter if nobody's at the other end ready to flip the switch. By May 2025, management admitted it expected to convert only about 5% of the 3,400-order backlog. By February 2026, Cogent quietly stopped reporting backlog numbers altogether when it released its Q4 results.

The actual revenue tells a more modest story. Wavelength revenue was $14.8 million in Q2 2026 — up 64% year-over-year, from a base of $9.1 million a year prior. Against total service revenue of $235.6 million, wavelength accounts for roughly 6% of the business. That's real growth, just not a growth engine.

Meanwhile, the core internet access and data transport business that Cogent built over two decades is shrinking. Service revenue fell 4.3% year-over-year in Q2 2026 to $235.6 million, and total customer connections declined 2.4% to 115,839. Off-net connections fell 12.2% year-over-year. The company is selectively exiting some legacy contracts to refocus on higher-margin services, which explains part of the churn — but the net effect is still a contracting revenue base.

This is where the balance sheet enters the picture and changes what the revenue numbers mean. Cogent acquired Sprint's wireline business for a price tag that loaded the company with secured debt. At the end of Q2 2026, the net leverage ratio was 6.23x — even after adjusting for amounts owed by T-Mobile. For context, a leverage ratio above 4x is generally considered heavy in telecom; above 6x, you're in "we need things to go right" territory.

The company's been using asset sales to manage the debt. In Q2 2026, Cogent sold ten owned data centers for $224.2 million and committed to use at least $175 million of the proceeds to repurchase its 2032 secured notes at a discount. Through July, it had bought back $138.8 million in par value of those notes, realizing a $13.4 million gain. It's a rational move when your own debt trades below par — but it's also a signal that the company is selling real estate to reduce its borrowing, not raising cash to grow the business.

EBITDA margins are a bright spot. Adjusted EBITDA hit $71.1 million in Q2 2026, a 30.2% margin, up from 29.3% in Q1. GAAP gross margin expanded to 24.5% from 13.6% a year earlier. The margin improvement comes from cost reductions, revenue mix shifts toward on-net services (which Cogent doesn't have to pay third parties to deliver), and the natural economics of a fixed-cost network that benefits from every dollar of revenue above the cost line. But margin expansion on declining revenue is a narrowing pie with a thicker slice — improving for Cogent, but not replacing top-line growth.

Then there's the dividend, which tells you what the board believed about cash generation over time, and what it had to change its mind about. Cogent paid a dividend for 13 consecutive years, increasing it each year. In November 2025, the company cut it by 98% — from $0.10 per share quarterly to $0.02. At the current stock price of around $9.60, the annualized dividend yield is about 0.8%. The forward yield looks artificially higher in some reports because of how the last full year's dividend is annualized, but the actual cash being paid out is $0.08 per share per year.

The dividend cut was a function of the same dynamics the lawsuit traces: the wavelength growth story wasn't materializing fast enough to support the legacy payout, the core business was shrinking, and the debt service was the priority. It's an honest read of the cash flow hierarchy — debt holders get paid first, then operations, then shareholders.

And speaking of shareholder alignment, the margin loan incident is worth noting. In August 2025, JPMorgan and Royal Bank of Canada seized and sold roughly $82.5 million of CCOI stock from CEO David Schaeffer after he couldn't meet a margin call on pledged shares. Cogent's SEC filings had previously stated there was "no material risk" of an involuntary CEO stock sale, based on Audit Committee review — while also warning in generic terms that pledged-share foreclosures could disrupt the market. The lawsuit says these disclosures were contradictory and inadequate. The practical point is simpler: when your CEO's personal finances are leveraged against your stock price, the incentive alignment you're supposed to get from executive ownership gets complicated by what happens when the stock falls. The stock was at $86 in late 2024. By the seizure, it was nowhere near there.

So what does Cogent Communications look like today, stripped of the backlog narrative?

It's a telecom company with a shrinking core internet business, a real but still-small wavelength growth opportunity, and a balance sheet that requires disciplined deleveraging. The market cap is about $490 million with an enterprise value of roughly $2 billion — meaning the debt is four times the equity value. The stock has fallen 55% year-to-date and 75% over the past twelve months.

The wavelength business is genuine. Cogent now offers optical services across 1,137 locations, and wavelength connections grew 66% year-over-year to 2,445. The addressable market is large and the incumbents are aging. Cogent's two-week provisioning cycle is a real competitive advantage — if customers actually want to accept the service. The company stopped reporting backlog, which is probably the right call. A metric that requires a disclaimer every quarter is a liability, not an asset.

The investment case hinges on whether the wavelength business can grow fast enough to offset core revenue declines and generate enough cash flow to bring the leverage ratio down from 6.23x to something sustainable. The data center sales show management is willing to monetize assets to de-lever, but that's a finite resource. Margin expansion helps, but it's margin expansion on a shrinking top line.

For an investor, the risk-reward here is straightforward. The stock has been punished severely, and the company has real fiber infrastructure, a growing wavelength franchise, and improving margins. But the debt is heavy, the core is contracting, the dividend is essentially gone, and the growth engine that was supposed to carry the transition is still too small. The lawsuit itself doesn't change the business — but the fact it exists tells you the company was building its investor narrative on a metric that couldn't survive contact with reality. That's a signal about how to think about the management's other forward-looking statements, regardless of whether the case is ever resolved.