Live: Complete Coverage of SoundHound AI (SOUN) Q3 Earnings

Here are three incremental takeaways from SoundHound AI’s conference call:

1) “Polaris continues to prove its superiority in accuracy, speed, and cost.”

Management emphasized that the company’s in-house multimodal foundation model is cutting error rates by “as much as three times” compared with third-party systems. The CEO added that Polaris can operate “in our cloud, in their cloud, on the edge, or on-prem,” giving enterprise clients flexible deployment choices and avoiding “the explosion in model costs others are battling today.” That combination of higher accuracy, lower compute cost, and hosting freedom positions SoundHound to expand margins as customers standardize on its stack rather than paying hyperscaler usage fees.

2) “Early agentic AI customers are seeing up to 10-fold improvement in containment rates.”

The CFO quantified performance gains from Amelia 7 migrations: 25 percent higher NPS, 15 percent higher customer satisfaction, and 35 percent less design effort. He added that roughly “75 percent of our customers” should move to Amelia 7 “by mid-next year.” Those metrics show real production-level lift, not just proof-of-concept hype, and suggest SoundHound’s “Agentic+” framework is resonating in complex enterprise workflows. The Interactions acquisition should accelerate that adoption by adding workflow-orchestration tools and giving SoundHound more opportunities to replace outside models with its own.

3) “We’ve already successfully placed live voice orders from cars… and look forward to seeing these go into full production in 2026.”

Voice Commerce, long a demo topic, is now moving toward mass deployment. Executives disclosed that SoundHound is “nearly fully integrated with two tech platform giants” to expose voice ordering to “many millions of users,” and in talks with “two prominent global TV manufacturers” for in-home ordering. Combined with the “double-digit-million” IoT device commitment from a Chinese robotics partner, these comments reveal a broad commercialization path across automotive, consumer electronics, and IoT—each with high-volume recurring potential.

SoundHound’s conference call will kick off at 5pm ET. Will post updates of the call after it starts.

Another look at the headlining numbers. SoundHound AI now down 2.39% after-hours. Which follows up the rough day in the market for most Nasdaq stocks, but SOUN ended the tradiing day down 9.48%.

CEO Keyvan Mohajer:

“Enterprise AI adoption is booming globally, and SoundHound is strengthening its leading position with deployments in millions of end points across highly diversified industries and customers. We see enormous potential in the near and long-term horizon, and we’re positioning our business to take full advantage.”

CFO Nitesh Sharan:

“Q3 was another strong mile marker on our journey to transform human-computer interfaces through natural language conversations and voice AI that makes technology more seamless and accessible. We continue to deliver high growth with a diverse customer base, expanding product suite, and a balance sheet that enables us to stay on the offense.”

Quarterly highlights on business units and outcomes.

• Agentic AI / Amelia 7.3 — integrates Speech-to-Meaning®, enhancing latency and conversational realism.

• Voice Commerce — adds restaurant reservations & parking payments; OEM launches expected 2026.

• IoT & Robotics — major China-based OEM deal for “double-digit millions” of AI-enabled devices.

• Automotive — expanding with Jeep (EU), Togg (Germany), Italian fleet OEM, and two Indian two-wheeler brands.

• Restaurants — national deployments at MOD Pizza, Habit Burger, Red Lobster, Torchy’s Tacos + franchise wins with Firehouse Subs and Five Guys.

• Healthcare & Finance — multiple renewals and new Amelia 7 contracts across hospitals, insurers, and banks.

• Channel Partners — new alliances with Telarus and VOXai to extend enterprise reach.

Stock bouncing back, just down 2.66% now. Here is a quarterly highlight of the numbers with my notes:

While the company did not issue formal Q4 guidance in the filing, management is expected to reaffirm its full-year 2025 revenue range of $160 – $178 million, consistent with its Q2 update.

After-Hours: ↓ 8.15 %

SoundHound shares are under pressure after the company’s Q3 print failed to clear high investor expectations. Revenue came in roughly in line with the Street but profitability moved sharply in the wrong direction, weighed down by one-time non-cash acquisition charges.

Earnings vs. Expectations

Topline growth remained strong at +68 % YoY, marking another quarter above $40 million. However, a $66 million fair-value adjustment on contingent acquisition liabilities drove a much wider net loss of $109 million, pushing EPS to –$0.27. Investors are reacting to that margin reset rather than the revenue beat.

SoundHound AI (Nasdaq: SOUN) is down close to 8% before earnings as it is  a proxy for overall sentiment in the artificial intelligence sector, making it highly sensitive to shifts in investor mood. With the recent sell-off in major AI names like NVIDIA, AMD, and Palantir, smaller-cap companies such as SOUN are being hit harder as traders rotate out of high-growth, speculative plays. The broader pullback stems from growing concerns about an “AI bubble” following comments from OpenAI’s CFO about potential government support and NVIDIA’s CEO suggesting that China could outpace the U.S. in AI development.

These headlines triggered fears that the AI boom may have run ahead of itself, prompting profit-taking across the space.

For SoundHound, whose valuation is built more on future potential than current profitability, that means volatility is amplified. Despite this, the company’s long-term story remains tied to the rise of voice and conversational AI applications in cars, devices, and enterprise systems.

Today’s early sell off is not as much about sellers getting out before the earnings, its more a reflection of the greater sell-off in AI stocks today.

The NASDAQ will close down 1.77% on the day.