51Talk (COE) Q2 2026 Earnings Call Transcript

51Talk reported Q2 2026 net revenue of $32.4 million, up 58.8% year over year, while gross billings rose 38.1% to $39.3 million and exceeded management's guidance; active students increased 51.3% to 138,100. The operating loss narrowed to $2.1 million from $3.2 million despite a 41.8% increase in operating expenses, driven by sales and marketing spending and a 96.6% increase in product-development investment. Management guided Q3 gross billings to $41 million-$43 million, implying 4.3%-9.4% sequential growth, and highlighted the July launch of its AI-powered Global Communicator learning product.

Analysis

COE’s key equity question is no longer top-line demand but whether customer-acquisition spend can be tapered without impairing enrollment. Student growth has lagged revenue growth only modestly, while marketing remains the dominant cost center; a credible reduction in sales-and-marketing intensity over the next two quarters could move the business to operating profitability disproportionately quickly because G&A is largely fixed. Conversely, teacher compensation and payment-processing costs are scaling at least as fast as revenue, limiting the value of gross-margin expansion unless the AI product demonstrably reduces content creation, support, or instructor-utilization costs.

The deferred-revenue balance creates favorable near-term liquidity and demand visibility, but it is not equivalent to unrestricted earnings power: it represents future service obligations and could become a refund risk if retention or lesson completion deteriorates. The most important 1-3 month catalyst is evidence that Q3 billing guidance converts into revenue while operating cash flow remains positive and marketing expense grows materially below billings. A higher-quality signal would be disclosure of repeat purchase, CAC payback, lesson frequency, and refund rates; absent these, management’s efficiency narrative remains unverified.

The AI/Oxford partnership should not command a software-style multiple yet. Content generation may lower marginal launch costs over 6-18 months, but AI-enabled language content is readily replicated by larger edtech platforms and could increase competitive promotional intensity rather than create pricing power. For a thinly traded small-cap ADR, liquidity, governance, FX/country exposure, and limited analyst coverage can dominate fundamental upside; this is an execution watchlist rather than a broad AI-theme proxy.

Key Decisions for Investors

• Maintain a watchlist-only stance on COE until the next results establish positive operating income or a clear sequential decline in sales-and-marketing as a percentage of billings. The required diligence inputs are CAC, retention/refund rates, and active-student monetization; without them, risk/reward cannot be reliably underwritten.
• If COE trades down more than 15% on a headline miss while Q3 billings remain within the stated range and operating cash flow stays positive, consider a small long with a 3-6 month horizon. Target is a rerating on demonstrated breakeven; exit if billings miss guidance or gross margin falls below 70%, signaling processing/teacher-cost pressure.
• Do not use NFLX or NVDA as sympathy longs: their inclusion is editorial and has no economic linkage to COE. The relevant comparison set is private/global online language-learning platforms, where AI lowers barriers to entry and therefore raises the burden of proof on COE’s retention and unit economics.