CEVA (CEVA) Q2 2026 Earnings Call Transcript

CEVA reported Q2 revenue of $29.0M, up 13% YoY (and 7% sequentially), with licensing and related revenue rising 21% YoY to $18.2M (highest in 3 years). Non-GAAP operating income increased to $3.1M and the non-GAAP operating margin expanded to 11% from 3% a year ago, alongside non-GAAP net income of $2.3M (+28% YoY) and EPS of $0.08. The company raised full-year 2026 revenue growth guidance to 13%–15% (from 12%) and expects non-GAAP operating income to rise ~70% YoY; Q3 revenue guidance is $30.5M–$34.5M as royalty momentum strengthens with wireless connectivity and automotive AI ramps.

Analysis

The important shift here is not the quarter itself, but the monetization model: CEVA is moving from selling isolated blocks of IP to embedding itself deeper into customer architectures. That usually shows up first as a better license fee, then months later as a higher royalty stream with less churn risk and more pricing power per socket. If that trend persists, the real losers are internal silicon teams and smaller point-solution IP vendors that cannot offer an integrated stack; the likely secondary beneficiary is any customer that can accelerate tape-out by buying rather than building, which supports faster device refresh cycles across wireless and edge AI.

The market should treat the AI contribution as a lead indicator, not a cash flow event. The path from design win to meaningful royalty is long enough that the next 1-3 quarters will mostly be about conviction in the pipeline, while the 6-18 month story depends on conversion and whether these wins scale beyond a handful of strategic customers. The biggest falsifier is any sign that H2 handset/memory constraints suppress volume or that the platform wins fail to expand into production on schedule; if that happens, the stock can re-rate back to a lumpy licensing multiple rather than a durable royalty compounder.

Contrarian view: consensus may be underestimating how much of this quarter was mix-driven and overestimating how quickly it becomes recurring. The stock can still work, but the cleaner thesis is not "AI enthusiasm," it is "higher content per design plus operating leverage." That argues for buying weakness tied to execution noise, not chasing a gap higher after a good call.

Key Decisions for Investors

• Long CEVA on pullbacks over the next 1-3 weeks; target a 6-12 month hold for the royalty conversion story, with downside thesis broken if Q3 revenue fails to hold the raised $30.5M-$34.5M range or if operating income stops expanding.
• Pair long CEVA / short SOXX to isolate the idiosyncratic rerating from platform-IP adoption; this is cleaner than a directional semiconductor bet if the market stays risk-on.
• Add to CEVA only after confirmation that the raised full-year guide is being driven by royalty conversion, not just one-off license fees; watch the next two prints for sequential royalty growth and any commentary on design-win-to-production timing.
• Treat the handset/memory shortage as a watch item, not a thesis killer; if low-end smartphone volume weakens materially in the next 1-2 quarters, reduce exposure because CEVA's near-term royalty leverage is more cyclical than the AI narrative suggests.