SoundHound Gets a Buy. Profitable AI Stock Cerence Gets a Hold. (CRNC)

Profitable AI stock rated Hold, SoundHound rated Buy

0:00 Cerence, ticker symbol CRNC, closed at $8.93, and despite reporting a profit, the average Wall Street analyst recommendation is still Hold. Meanwhile, SoundHound, another voice AI company, is losing money, but gets a Buy rating with analyst price targets more than 160% above its stock price.

What Cerence actually does: voice AI in your car

0:18 So why are analysts more excited about the company losing money? You know when you ask your car to call somebody, it picks the wrong person, and you end up reaching for your phone because that would have been easier in the first place? Imagine asking for a coffee shop that's still open and on your way home without breaking that into three separate commands. That's the kind of experience Cerence is trying to make possible. Here's the business. Cerence makes software that helps your car understand what you're saying, so you can ask for directions or find a gas station without pulling over to type.

0:45 Things like measuring your emotions, using the cameras in the car to read street signs and billboards as you're driving. You pass a billboard, you can ask the car, hey, what was that billboard that I just passed? The car will have read it. But it's bigger than it sounds, because on the May earnings call, Cerence's CFO said about 50% of the cars produced worldwide included Cerence's technology. The clearest example is Volkswagen, which said in June 2024 that it was adding ChatGPT to many of its models through Cerence, letting drivers ask questions in a more natural way.

1:15 But answering your questions and driving your car are very different tasks. So when you hear AI in this video, think voice assistant, not robotaxi. So the newer deals are the interesting part.

BYD, Stellantis and 5 of 6 xUI wins

1:24 In April, BYD picked Cerence's newest platform, called xUI, starting with one model, the ATTO 2 DM-i. And in August, Cerence announced a multi-brand xUI award from Stellantis, the company behind Jeep, Ram and Chrysler. CEO Brian Krzanich, who used to run Intel, put a number on it during the May call: by his count, Cerence had won five of the six xUI bids thus far. So BYD and Stellantis have both selected Cerence's software, but selecting it is only the first step, because the rollout and the contract terms determine how much revenue those contracts actually end up producing.

1:59 We hit it out of the park on almost every single parameter in the third quarter. Which brings us to the money, because Cerence has actually produced a profit, but we need to understand where that profit came from.

The profit: GAAP net income and cash flow

2:10 For the quarter that ended June 30th, 2026, Cerence reported $1.5 million of GAAP net income, which means it earned a profit using standard accounting rules, rather than an adjusted measure that leaves certain expenses out. It also generated about $20 million of operating cash flow, meaning cash from running the business, down from about $23.7 million a year earlier. Management expects $81 million to $85 million for the full fiscal year, while the stock market values all of its shares at about $404 million.

2:38 SoundHound, by comparison, lost about $137 million over its last 12 months on about $203 million of revenue, and it's the one with the Buy ratings. So why does the profitable one of these two competitors get the Hold rating?

Concern 1: the one-time license check

2:52 Part of that answer is hiding inside that same quarter, because revenue was $69.6 million, up 12% from a year earlier, but $12.5 million of it was from one big license check that was not there at all the year before. Now here's why that really matters: the total only grew by about $7.4 million. So that means if you take that one check out, it actually shrank. Think about getting a one-time bonus at work. Your bank account looks better that month, but you wouldn't assume your regular paycheck just got bigger, and we need to make the same distinction here.

3:22 So the next question is how much revenue Cerence can bring in without another big license payment, and we already have a preview of that one. Management's forecast for the quarter ending September 30th is $61.1 million to $65.1 million of revenue, with no license deals expected, so a smaller total wouldn't automatically mean the business got worse. The encouraging part is connected services, which brings in revenue from services used after the car is sold, and that revenue grew more than 20% from a year earlier.

3:49 Except when an analyst asked the CEO on the May call whether that growth was coming from the new xUI platform, he was refreshingly honest about it. At that point, the new xUI platform wasn't actually driving the growth. So the question is whether those contract wins actually turn into cars using the software, which would mean additional revenue growth. That's why repeat business matters here. Just like you would rather cover your bills with a regular paycheck than depend on a bonus, Cerence needs revenue it can keep earning after a big deal closes.

Concern 2: convertible notes and dilution

4:18 The second concern is debt, because at June 30th, 2026, Cerence had about $127.6 million in cash and about $180 million of debt due July 1st, 2028. Imagine lending money to a business with an agreement that lets the debt turn into an ownership stake under certain conditions. You could end up owning part of the company instead of simply getting your loan repaid. That's the basic idea behind a convertible note, and if the company issues new shares to settle the debt, existing shareholders can end up owning a smaller percentage of the business.

4:45 In August, Cerence agreed to buy back $10 million of those notes at 92.75 cents for every dollar owed, which would leave $170 million if the deal closes, essentially paying off that portion of its debt at a discount. But the buyback needs careful consideration: just because Cerence is authorized to buy up to $30 million of its stock over the next 12 months doesn't mean that it will actually follow through with that. That matters because the average share count in the June quarter rose to about 45.2 million, from about 43.3 million a year earlier, so shareholders are dividing ownership among more shares.

Concern 3: automakers hold the pricing power

5:19 The third concern is that Cerence sells to automakers, which can compare competing software and negotiate the price, rather than to drivers, who would choose it directly. Think about buying a car. You might like its voice assistant, but you buy the whole vehicle from the carmaker, and that carmaker negotiates what Cerence gets paid for its software. And even after an automaker chooses Cerence, its vehicle production schedule helps determine when that software actually reaches drivers and starts contributing revenue.

5:42 So let's look at the ratings themselves, because they are not as uniform as Hold sounds.

What the analyst ratings actually say

5:46 TD Cowen is a Hold at $12. Goldman Sachs is a Neutral at $9. Craig-Hallum is a Hold, Weiss Ratings is a Sell by letter grade, and Cantor Fitzgerald is the one Buy, at an $11 price target. The average price target is about $10.67, which is about 19% above that last closing price, compared to SoundHound, which has about 160% of upside from the analysts. So clearly analysts are seeing more growth potential in SoundHound, even though Cerence is the one actually turning a profit currently.

Two kinds of investors

6:16 Now, there are two types of investors in this situation. One of them sees the large gap between the price target and the current price on the market and immediately rushes to buy that stock. The other one just wants to buy the profitable company, but I think both of them have a lot to consider here. A worthwhile step is checking whether that progress repeats.

3 checks for the next earnings report

6:34 So here are the three checks I would run on the next report, which, going by last year, should land around late November. The first check is whether the BYD and Stellantis deals are moving into production, with the software reaching actual cars and eventually contributing revenue. The second check is revenue apart from the big license payments, because we need to see whether the rest of the business is growing, especially connected services. The third check is whether Cerence actually buys back shares and reduces its debt, and whether those repurchases are enough to bring the total share count down.

7:02 If the rollouts stall or connected services growth slows, that would support the analysts' caution. But if repeat revenue keeps growing and the share count falls, the case for Cerence gets stronger.

Is the Hold right?

7:12 Nobody can predict which way that actually goes, and an analyst Hold rating is just their opinion. It's not a guarantee that the stock is actually going to stay right where it is. In fact, a lot of times being a contrarian will get you lower prices than just going with what the analysts are saying. So the question is whether Cerence can keep making its software useful to customers, and whether that can actually turn into repeat revenue and profits, because a single profitable quarter does not really answer that for us.

7:38 So what do you think? Do you think that Wall Street is completely missing the story here? Do you think Cerence is actually a Buy? Tell me in the comments. YouTube already picked your next video, and unlike Wall Street, it actually commits to a rating.