Agora, Inc. Sees Growth in AI and Real-Time Engagement

The headline numbers from Agora, Inc. tell a story of a company finding its footing in a challenging market. Revenue of $40.4 million for the second quarter of 2026 represents a solid 18% year-over-year increase. Net income is a positive $2.2 million, marking what CEO Tony Zhao proudly calls a seventh consecutive quarter of GAAP profitability. On the surface, it’s a picture of steady, disciplined growth. But the real story, the one that defines Agora’s future, is buried in the cost structure and a single, telling metric: the gross margin.

For a company sitting on over $360 million in cash and equivalents, the profitability is as much a function of financial engineering as operational prowess. A full $3.4 million of that quarterly profit came from interest income on its massive war chest. Strip that out, and the core business looks far less robust. The operating loss, while improved, still clocked in at nearly $1 million. This isn’t necessarily a critique; it’s the reality of a firm in transition. Agora is navigating the costly pivot from being a pure-play real-time engagement (RTE) infrastructure provider to a contender in the brutally expensive conversational AI arena.

This pivot is etched into the financials. While revenues grew 18%, the cost to generate those revenues—the cost of revenues—jumped nearly 29%. The culprit? The company explicitly cites “increases in bandwidth and server costs and costs related to conversational AI products.” The result is a gross margin squeeze, falling to 63.7% from 66.8% a year ago. Agora admits its AI products remain at a “sub-scale stage.” In plain English, they are currently expensive to deliver and likely sold at competitive, lower-margin rates to gain market share. Every dollar of new AI revenue is costing them more than a dollar of their legacy RTE business.

This is the essential gamble. The legacy business, powering video and voice for sectors like live shopping and telehealth, is the reliable engine. It’s what gives them the 104% dollar-based net retention rate, indicating existing customers are spending slightly more. But customer growth has nearly stalled, with Active Customers ticking up a mere 0.4% year-over-year. The future growth narrative, therefore, is entirely tied to AI. Zhao’s statement about AI agents beginning to “match or even surpass human performance” in tasks like market surveying and customer service is the hook. They are betting that this technology will trigger an “industry shift toward AI-led call center workflows,” a market vast enough to justify the current margin compression.

The market’s patience for this story will be tested by the balance sheet discipline. Agora is clearly trying to have it both ways: funding its AI ambitions while returning capital to shareholders. Their aggressive share repurchase program has retired nearly 25% of their shares since inception, a $159.9 million deployment that directly supports earnings per share figures. Simultaneously, R&D investment is up over 10%, almost entirely funneled into conversational AI. The company is walking a tightrope, funding a speculative, cash-intensive future while using its balance sheet to manufacture stability and shareholder returns today.

The guidance for next quarter—revenue between $41 and $42 million—suggests more of the same: steady top-line growth in the high teens, continued investment, and profitability sustained by its interest income. For investors, the question isn’t about this quarter’s earnings. It’s about the timeline. How long will the “sub-scale” phase last? When will AI revenue achieve the margins of the RTE business? Can they grow the AI pie fast enough to offset the margin dilution and reignite meaningful customer acquisition?

• Stable revenue growth
• Operating loss reduced
• Investment in conversational AI
• Share repurchase program
• Customer growth stagnation
• Future tied to AI revenue

Agora’s second-quarter results are a financial snapshot of a company caught between two identities. It is a profitable, cash-rich infrastructure provider patiently—and expensively—building what it hopes will be a high-growth AI software future. The numbers show the strain of that construction. The coming quarters will reveal if the foundation is being laid for a skyscraper or if the costs of building will remain a perpetual drag on the solid, but unspectacular, business that already exists.