Radware has $460 million sitting in cash. Its market cap is $1.18 billion. That means roughly 39 cents of every dollar you pay today is just cash on the balance sheet — before you get any of the cybersecurity business, the cloud ARR growing at 23% annually, or the three new AI security products the company launched in a six-week window at the start of 2026.
The stock has gained 9% over the past 12 months. Analyst consensus is “Hold.” Most retail investors have never heard of it.
That’s the setup.
The Numbers: Record Everything, Again
On February 11, 2026, Radware reported Q4 and full-year 2025 results that hit records across every metric that matters:
- Q4 revenue: $80.25M — beat the $78.65M consensus by $1.6M, up 9.9% year-over-year
- FY 2025 revenue: $301.9M — first $300M year in company history, up 10% YoY
- Q4 non-GAAP EPS: $0.32 — beat the $0.30 estimate, up 19% from $0.27 in Q4 2024
- FY 2025 non-GAAP EPS: $1.15 — up 32% from $0.87 in 2024 (record)
- FY 2025 GAAP EPS: $0.45 — up from $0.14 in 2024 (221% improvement)
- FY 2025 operating cash flow: $50.1M
- RPO (remaining performance obligations): $400M — record, represents forward revenue pipeline
But the numbers that matter most for long-term value are the recurring revenue metrics. Cloud ARR hit $95.2M in Q4, up 23% year-over-year and 7% sequentially — one strong quarter away from crossing $100M. Total ARR rose to $251.0M (+11% YoY), and subscription revenue grew 21% for the year. This business is quietly transitioning from hardware appliances to a subscription-first model, and the ARR trajectory tells you how far that shift has progressed.
The Cash Story Nobody Is Telling
Radware has been debt-free for five years. As of December 31, 2025, the company held $460.6 million in cash, deposits, and marketable securities against a $1.18 billion market cap. Strip out the cash and you’re paying roughly $720 million for a business that generated $50.1 million in operating cash flow in 2025 — a cash-adjusted P/OCF of about 14x.
On February 13, 2026, Radware’s board authorized a new $80 million share repurchase program (the “2026 Plan”) running through March 15, 2027. That’s 6.8% of the current market cap. Management doesn’t do buybacks when they think the stock is fairly valued — they do them when they think the stock is cheap.
The combination of a strong cash position, no debt, and a buyback program gives this company more financial flexibility than nearly any small cap at this size. If the cloud ARR growth story plays out, the cash buffer prevents any near-term financial distress. If it doesn’t, there’s still $460M of hard assets behind the stock.
The AI Security Buildout: Three New Products in Six Weeks
Here’s what Radware did between January 8 and February 3, 2026:
January 8: Radware’s threat research team published the discovery of “ZombieAgent” — a zero-click, indirect prompt injection vulnerability targeting AI agents, specifically demonstrated against OpenAI’s Deep Research agent. The attack requires no direct user interaction; it lets attackers hijack AI agents that interact with corporate systems, emails, and workflows. The research paper landed on security news sites before Radware had a product to sell. That sequencing matters: discover the threat publicly, build the product, own the narrative.
January 20: Radware launched its end-to-end API Security Service — runtime protection, posture management, and business logic protection for APIs. APIs have quietly become the primary attack surface for enterprise systems. Radware’s own 2026 Global Threat Analysis Report (released February 19) showed application and API attacks up 128% year-over-year in 2025.
January 26: Radware acquired Pynt, a 15-person Israeli API security testing startup, for “tens of millions” of dollars. Pynt covered the development phase — finding API vulnerabilities before they hit production. Combined with the API Security Service (runtime defense), the acquisition gives Radware the full lifecycle: test in dev, protect in production, respond to attacks in real time.
February 3: Radware formally launched its Agentic AI Protection solution, the commercial product that came out of the ZombieAgent research. Designed to protect enterprises as AI agents proliferate across their infrastructure.
The threat data in Radware’s own research supports the product strategy: network DDoS attacks were up 168.2% year-over-year in 2025. Web DDoS was up 101.4%. Bad bot activity up 91.8%. The company doubled its global attack mitigation capacity to 30 terabits per second in January 2026, and the demand data from their own threat report justifies why.
This isn’t a company that launched AI products because it’s trendy. Radware’s core business — defending APIs, web applications, and infrastructure from active attacks — is the exact thing AI agents break when they’re deployed poorly. The product roadmap follows the threat landscape.
Why Analysts Are Still Saying Hold
The skeptic case is real and worth taking seriously.
The GAAP P/E is 72.9x. That’s not cheap for a company with a Hold consensus and limited sell-side enthusiasm. Two analysts rate it Buy, two Hold, one Sell. The average price target is $30.00 — roughly 8-9% above current prices. Jefferies sits at $25.00. You’re not getting a deep value setup here on traditional metrics.
Q1 2026 guidance shows a sequential revenue step-down: $78-79M, against Q4’s $80.25M. Management cited higher planned OPEX investments (including a ~$1.5M FX headwind) as a margin pressure. That’s not alarming in isolation, but guidance that calls for lower sequential revenue in the quarter immediately following a beat doesn’t give traders a reason to buy.
Regional softness showed up in Q4: Americas revenue fell 4% year-over-year despite strong bookings, and APAC was down 3%. Management attributed the gap to timing — deals booked late in the quarter that hadn’t converted to revenue recognition yet. One quarter of timing noise is fine. Two consecutive quarters would be a different conversation.
And there’s the stock’s historical behavior: over the last five earnings cycles, RDWR’s average post-earnings next-day move was -0.98%. The company beats expectations consistently and the stock consistently goes nowhere. That pattern either reflects rational market pricing (stock was already fairly valued going in) or it reflects a structural lack of institutional sponsorship for a small Israeli cybersecurity company that doesn’t get much analyst coverage. Probably both.
Also worth flagging: Radware had an Investor Day on February 17, 2026, in New York City. The stock closed the following day. Watch the transcript from that event — if management offered anything substantive on cloud ARR cadence or Agentic AI Protection pipeline, it’ll matter more than any single quarterly number.
The Comparable That Helps Frame This
We’ve covered Fastly (FSLY) and Amplitude (AMPL) here as AI-adjacent small caps trading at discounts to their growth. Radware is different: it’s not a platform company hoping AI adoption lifts its metrics. It’s a security infrastructure company whose TAM expands directly as AI agents proliferate and create new attack surfaces.
For another angle on the fraud and security infrastructure space, see our coverage of Mitek Systems (MITK), which rides the AI fraud wave from the identity verification side. Radware and Mitek are solving adjacent problems in a world where AI is both a productivity tool and a new category of threat vector.
The Verdict
At current prices (~$27-28), Radware is a good business at a fair price. Not a screaming buy.
The ex-cash valuation (~$720M enterprise value against $50M+ annual operating cash flow) is reasonable for a company growing cloud ARR at 23%. The $400M RPO gives forward revenue visibility most small caps can’t match. The AI security product pipeline — ZombieAgent research, Agentic AI Protection, the Pynt acquisition, the API Security Service — positions Radware ahead of a threat landscape its own data confirms is getting worse fast.
But the Q1 revenue guidance step-down, the 73x GAAP P/E, and the stubborn Hold consensus mean the near-term catalysts aren’t obvious. The stock needs either cloud ARR crossing $100M (one quarter away) or early evidence of Agentic AI Protection deals closing.
A better entry is $24-25, near the 50-day moving average ($24.34). At that level, the cash-adjusted valuation becomes genuinely compelling and you’re buying closer to what the bears think the stock is worth. At $27-28, you’re paying a reasonable price for a business on the right side of an important trend — but you’ll need patience while the market catches up.
Watch Q1 2026 results closely. If Americas revenue bounces back and cloud ARR hits $100M, the stock’s re-rating case gets a lot simpler.
This is not financial advice. I hold no position in RDWR.
Sources: Radware Q4/FY2025 Earnings Release — StockTitan | Radware Acquires Pynt — SC Media