Radcom (RDCM) Q3 2026 earnings review

Radcom returned to profit, but delayed projects wait until 2027

Radcom, which sells software that helps phone companies check their network quality, climbed back from a weak spring. Sales rose 19% from the second quarter but stayed 24% below a year ago. Operating profit excluding one-offs returned, though lower spending supplied about half of the gain. Chief executive Benny Eppstein kept the full-year plan, which by our math needs about $16 million of sales in the fourth quarter.

At a glance
Revenue$14.0M −23.6% from a year ago; +19.4% from last quarter
Operating profit, excluding one-offs$0.7M loss of $2.2M last quarter
Cash and deposits$111.4M no debt; $109.7M three months earlier
Full-year sales outlookUnchanged $57–63M; middle of the range: 16% decline

⚖️ Verdict: ⚪ Neutral

The story is unchanged, because the stronger quarter matched a plan that already counted on it. The delayed network projects still point to early next year, though in July Eppstein had also allowed for a fourth-quarter start. One plus: two operators replaced rival vendors with Radcom.

The question now is whether operators restart the delayed projects once they set next year's budgets. If server prices ease, the work resumes and double-digit growth returns; if they stay high, the projects slip again. The outlook Radcom sets for next year will tell.

🐂 Bull Case

🟢 strengthening COMPETITION

Two Operators Swapped Their Old Supplier for Radcom

Radcom sells service assurance, software that shows phone operators how their network performs for each subscriber. Chief executive Benny Eppstein said the quarter brought two new customers, VNPT and CETIN Networks, "each replacing the incumbent vendor through a competitive process".

VNPT is one of Vietnam's largest operators. CETIN Networks, part of a group with more than 12 million customers, already runs RADCOM ACE, the Israeli firm's main platform, in production.

Radcom argues that costly servers help it, because its software runs on hardware operators already own. But neither contract carries a disclosed value, so the wins cannot yet show whether they offset the delayed work at larger customers.

What to watch: whether VNPT widens the deal to its whole network, which Radcom says it is working on. That would show the first contract opened a larger account.

🟢 strengthening PRODUCT

Radcom Released Its ADM Analytics Tool on Time

Three months ago Radcom promised that ADM, a tool that lets operators' own teams build network reports without vendor help, would reach all customers this quarter. The slides now list ADM as available, so the Tel Aviv group kept its date.

ADM feeds Neura, Radcom's AI agents that search for the root cause of network faults. The slides say "Several customers and prospects are now evaluating Neura".

Chief executive Benny Eppstein said Radcom is investing in "our agentic AI roadmap to drive growth". But the company has reported no revenue from Neura or ADM, so these products do not yet support the sales outlook.

What to watch: a first paid Neura contract, or any revenue figure for it, in the fourth-quarter results. A paid contract would show that operators will pay extra for the AI tools.

🐻 Bear Case

🔴🔴 persistent MACRO

Delayed Customer Projects Still Wait for Early 2027

Several large operators postponed planned network expansions this year because the servers they need became expensive. Radcom's slides say other vendors in its industry "report the same delays". In July, chief executive Benny Eppstein allowed that one or more projects might restart in the fourth quarter.

This quarter Eppstein pointed only to "positive signs that some deployments delayed in 2026 will begin to move forward in early 2027". Customers have not started paying ahead for that work either. Prepayments and deferred revenue, money received before Radcom delivers, totalled $1.1 million, against $3.1 million a year ago.

This matters because, at the middle of its range, the full-year plan needs the fourth quarter to outsell the third without these projects. If server prices stay high when operators set next year's budgets, the restart could slip again. Then the promised double-digit growth for 2027 would rest on new customers alone.

What to watch: whether the fourth-quarter results name a restarted project, alongside the first outlook for 2027. A named restart would show the delay is ending on the timetable Eppstein set.

🔴 persistent MARGIN contradicts narrative

Half the Profit Gain Came From Lower Spending

Radcom's headline is a return to operating profit excluding one-offs such as stock pay. Spending on that basis dropped 13% from the second quarter, with sales and marketing down 25%. By our math, lower costs supplied about half of the improvement.

  • Research and development: 35.0% of sales, up 10.5 percentage points from a year ago
  • Cost of delivering the products: 26.2% of sales, up 3.3 points

Compared with a year ago, costs still take a bigger share of every dollar of sales. The Israeli firm has not explained the drop in sales spending, so it may not last.

What to watch: sales and marketing spending in the fourth quarter. A return to the $4.2–4.7 million a quarter of the past year would leave profit resting on a faster sales recovery.

🔴 fading CAPITAL ALLOCATION

The $20 Million Buyback Has Not Started Yet

In September Radcom's board approved buying back up to $20 million of shares over 18 months. That sits at the bottom of the range chief executive Benny Eppstein described in July. The slides say the "Creditor objection period under Israeli law ended with no objections", so the legal step is done.

The company bought no shares in the quarter. Eppstein said purchases would start "after the current earnings-related blackout period ends", and he again said Radcom will "remain open to partnerships and targeted acquisitions". For five quarters before July, management named acquisitions its first use of cash and made no deal.

What to watch: shares repurchased in the fourth quarter. Buying about $3.3 million a quarter would match the plan's pace, and the $111.4 million in cash and deposits covers it easily.

🔴 persistent DISCLOSURE

Risks this quarter didn't answer

Four standing concerns got no new figures in the release or the slides, so each stays where it stood three months ago.

  • Customer concentration: Radcom still names AT&T and Rakuten Mobile as key accounts without giving their share of sales; the annual report's customer table will show it.
  • Paying for AI: Radcom has set no public price model and reports no revenue for its AI agents; a first priced contract would move this.
  • Shekel costs: the currency hedge on Israeli-shekel spending runs only to the end of 2026; the cost plan for 2027 will show the exposure.
  • New board: the directors who took over in May have made one large decision, the buyback; any change to strategy or management would test them.

💲 Other KPIs

Trade receivables (26Q3) $19.7 million
⇄ reversing

Money customers owe Radcom rose 14.1% from a year ago, while sales went the other way. The balance equals 141% of a quarter's sales, up from 94% a year ago. That ratio eased after peaking at 183% in the spring, so collections have started to catch up.

Cash generated in the quarter $1.7 million
⇒ stable

The group added cash for a second straight quarter despite weaker sales. Its own measure, the change in cash and deposits, improved from $1.3 million in the spring but stayed far below $5.1 million a year ago. Interest income of about $1.5 million helped.

Diluted share count (26Q3) 17.3 million
⇒ stable

The share count kept edging up, 2.2% above a year ago, because Radcom pays staff partly in stock. That stock pay ran at $1.5 million this quarter. The buyback approved in September is the first tool that could reverse the drift.

🔮 Guidance

FY2026 Revenue $57–63 million
🠆 unchanged from $57–63 million (set July 30, 2026)
⇗ accelerating

Unchanged. The plan still has sales falling about 16% this year. By our math, it leaves about $15.6 million for the fourth quarter. That is 11% above the third quarter, yet 17% below a year ago. The low end needs only $12.6 million, so a second cut since July would take a drop from the third quarter's level.

FY2026 Profit, excluding one-offs Profitable on a non-GAAP basis
🠆 unchanged from Remain profitable on a non-GAAP basis (July 2026)
⇒ stable

Unchanged. Radcom still expects a profit for the year excluding one-offs. After nine months that profit stands at $5.1 million, so only a fourth-quarter loss of that size would break the plan. The weakest quarter this year lost $1.5 million. The Israeli group has not said whether the target means operating profit or net profit, which includes interest.

H2 2026 Free cash flow Positive for the second half
🠆 unchanged from Aim to remain free cash flow positive in the second half (July 2026)
⇒ stable

Unchanged. Radcom aims to generate cash across the second half. The third quarter added cash, so the target looks within reach. One catch: the firm measures cash flow as the change in cash and deposits. Share purchases after the blackout will lower that figure, and the company has not said whether it will exclude them.

FY2027 Revenue growth Double-digit growth (10% or more)
🠆 unchanged from Return to double-digit percentage revenue growth in 2027 (July 2026)
⇗ accelerating

Unchanged. Radcom still plans to return to double-digit sales growth next year. By our math, that means at least $62.7–69.3 million, depending on where this year lands. Even the top of that range would leave sales short of the 2025 level. The forecast leans on delayed projects starting in early 2027, as chief executive Benny Eppstein now signals.

❓ Key Questions

How large are the delayed projects, and whose are they?

Radcom has never given the value or the customers behind the postponed expansions. Without them, readers cannot size the recovery that next year's growth plan depends on, or tell whether AT&T is among them.

Why did sales and marketing spending drop by a quarter?

Sales spending fell about $1.2 million from the second quarter, and staff-pay accruals on the balance sheet also dropped. If lower bonuses or one-off reversals drove it, the return to profit may not hold.

Could any delayed project still start before year-end?

In July Eppstein allowed for a fourth-quarter start; the latest release and slides mention only early 2027. The answer shows whether the fourth quarter must reach the middle of the plan without that work.

Will share buybacks count against the cash flow target?

Radcom measures cash flow as the change in its cash and deposits, so purchases lower it directly. Readers need to know whether the second-half cash target leaves the buyback out.

Has Neura, ADM or any partnership produced revenue yet?

Radcom says several customers are evaluating Neura and that ADM is live, but it reports no revenue from either, or from its NVIDIA and ServiceNow work. That shows how much next year's plan depends on paid AI tools.