SoundHound - AI for voice systems

Investment Analysis Clubs Saul’s Investing Discussions
1

Doing some research into any new stock ideas that could match the investment ideas of Sauldom, I came up with SoundHound. I am posting this for feedback on how my analysis stands up, and feedback on how to better tune what I need to look at going forward.

Yep, SoundHound started as that phone app that could recognize music in public and then tell you what song it was. I had that app on my phones for some time in the early 2010’s.

It has evolved into what looks like an AI powered, voice recognition powerhouse because of that history. They have all the data from all their years processing sound queries. Ala Tesla’s inhouse data, they have the same thing related to voice commands across their product lines. To me, they appear to be the best situated to deal with interactive response AI just based on the data and the history of their niche.

Starting the obvious negatives, that probably stop most from investing, we have a company that just started trading on the NASDAQ in April '22. They have income measured in the $milion$ and a total market cap of $550 mil as I write this. There are some customer concentration risks, they call out that some auto companies are over 10% of revenue. Lastly, they are still losing money to the tune of -$20 mil on revenue of $13 mil.

But, they do seem to be actively addressing all those things. It is just that they are relatively new to their growth and it is going to take time for them to balance all the things. Here are my notes.

Corporate losses - They started a restructuring effort in Q1 of this year to address the high losses and reduce overall operating expenses. As of 3rd Q they have increased revenue while dropping losses by 35% in one year. This looks like a great trend that I will keep an eye on.

Corporate revenue - Their revenue is tiny compared to other stocks we own, and it is very lumpy. However, they are predicting more growth for 4th Q and they expect to be break even by next year. (I couldn’t find specifics, just noted they said it in call transcripts.)

Corporate margins - This is where the stock popped up in my searches, they have margins over 70% for all the years I could find. Their revenue growth was big, but they are also just starting out so it is easy to get growth percentages to look good. They did drop off this quarter by a lot, 56% - 42% - 19%. The super low 19% is a YOY comparison, and it seems that Q3 '22 was a big year due to some contract revenue recognition, so it’s not like the business just failed in past quarter, they had tough comps due to being small enough that one contract impacted the numbers a lot.

Customers - (Automotive) They show over 20 different automobile manufactures as customers, across the globe, so they are pretty imbedded in that market. (Customer Service) They have just (my assumption from call notes) pushing into the restaurant and customer service world with AI voice interaction. They have specific restaurants they are trialing with (White Castle, Jersey Mikes, Krispy Kreme) but they are also working with a hardware partner for POS stations that means exposure to Toast, Square, Oracle customers using point of sale devices. (IOT Devices) They are also the voice response system for many household devices like TV’s (Visio), OEMs like Qualcomm and Motorola, and services like Pandora.

Does this outline look interesting to anyone else? It is AI, it is a newishly traded company that has been around for a while, it appears to be landing customers quickly, and I feel like the TAM here could be very big. As small as they are now, a buyout is HIGHLY likely in my experience, but I’d still call that a win. To me, this is NOT a story stock as they are executing, and they have actually added products. They are unique with their own datasets, just like some of the others we praise on the board.

What have I missed that needs to be considered? Let me know how to be more critical of these kinds of companies.

Lastly, here is my spreadsheet.

Customer count
1-'22 2-'22 3-'22 4-'22 1-'23 2-'23 3-'23
automotive >20
restaurant >3
IOT/other
ARR – YOY
4-'21 1-'22 2-'22 3-'22 4-'22 1-'23 2-'23 3-'23
$10,000.00 $25,000.00
Revenue - $Thous
4-'21 1-'22 2-'22 3-'22 4-'22 1-'23 2-'23 3-'23
$5,151.00 $4,290.00 $6,152.00 $11,186.00 $9,501.00 $6,707.00 $8,751.00 $13,268.00
-$21,847.00 -$25,103.00 -$30,668.00 -$30,061.00 -$30,680.00 -$26,369.00 -$21,932.00 -$20,197.00
Revenue Growth – YOY
4-'21 1-'22 2-'22 3-'22 4-'22 1-'23 2-'23 3-'23
15.00% 43.00% 178.00% 84.00% 56.00% 42.00% 19.00%
Gross Margins
4-'21 1-'22 2-'22 3-'22 4-'22 1-'23 2-'23 3-'23
77.00% 71.00% 71.00% 79.00% 73.00%
6 Likes
2

Two extra notes that I wanted to call out here.

Why is this not a story stock? They are already executing and rolling services out to customers. They are also innovating and they have already completed “phase 1” of a new solution for AI voice recognition:

I’m pleased to say we’ve completed the first phase of this project and successfully built a multilingual model that can handle queries in multiple languages interchangeably and respond with high accuracy with streaming audio in real time. We will be rolling out the first version of Polaris to our customers in the coming weeks. This is an important milestone, and we expect even more disruptive features in future phases of Polaris. We are excited about the potential applications of a model that can respond instantly and seamlessly and in different modalities to request in various languages across a huge range of potential queries. The commercial possibilities are clearly huge.

Another bit of kudos to track is that the company is ranked in the top 10 machine learning by Technology Magazine in a market that is set at $2 trillion TAM by 2030.

10 Likes
3

@dlbuffy, Hmm, I have SoundHound on my phone. It wasn’t free. I don’t recall exactly what I paid for it, maybe $2.00. Anyway, I bought it because the main (or maybe only) function it provided was the ability to provide a song title, artist name, etc. just by listening to it for a few seconds.

Later I installed Shazam. It does the same thing, only better (it will recognize songs that Sound Hound doesn’t recognize). And Shazam is a free app.

If you look at SoundHound in the Google Play Store you will see that it has recently garnered a lot of one and two star ratings (it used to get a lot of five stars). The main complaint is that it now fails to recognize a lot of songs. Users aren’t happy. Personally, I wouldn’t know as I’ve not used the app for years. I exclusively use Shazam for song recognition.

I’m not sure how that plays into the narrative you’ve just posted. You mentioned its origin as a song recognition app, but didn’t elaborate. Looks as if they’re changing direction, but if so they’ve done a terrible job of informing their current customer base. Maybe that isn’t relevant?

5 Likes
4

TMF is not super excited by this stock:
SoundHound AI Is Singing a More Pleasant Tune, but Does That Mean Investors Should Buy the Stock? | The Motley Fool

Debt, and share dilution are biggies here.

5 Likes
5

@brittlerock, it seems from dlbuffy’s post that the app was how they started but is not core to their future. They are aiming to sell sound AI capabilities B2B, which is of no interest to app consumers, so I wouldn’t expect them to have some proactive communication about it. The deterioration of the app’s quality may reflect that change of focus. I doubt it harms their brand with these B2B conversations.

@dlbuffy my big concern would be what is their moat? Yeah they have a large library of real-world audio clips, but how far will that take them? Probably the most direct application would be in very noisy environments, which is unlike the situation in a car (though would be useful in restaurants). Moreover, the most important aspect of the use case is likely not to be the voice recognition itself, but the intelligence to do the right thing in response. At both ends of that interaction, you have giants like MSFT, GOOG, AMZN, AAPL who have spent many years developing voice assistants, delivering advanced tech B2B, or both. I have a hard time seeing SOUN providing compelling value as a point solution, especially as state-of-the-art AI becomes more and more expensive and best accessed through major providers like MSFT or GOOG.

I feel this is reflected in their low revenue. They really have not attracted many customers. Most are in auto, which is 1) the environment I think they have the lowest advantage from their 1P data, 2) relatively small in terms of the number of customers they might eventually get, and 3) perhaps very suited to displacement by larger technology partners who can handle the full range of the auto makers’ needs. How many of these customers are really taking a bet on SOUN tech vs just satisfying one need in their systems in a way that could be swapped out at any time?

And one more negative take on the revenue. In 2022 they grew revenue 50%. Based on Q4 guidance, they are set to roughly grow 50% again. If they continue growing at 50% for three more years, they will finally eclipse their current operating expenses. If they don’t have a serious growth inflection (story stock) or dial their expenses way back (very unlikely), this isn’t a sustainable business at any kind of attractive profit margin. At the end of 2022 they were down to less than $10M in cash and equivalents, and in the early months of 2023 took on $25M in preferred equity financing and $100M in debt financing. This is the source of their dilution, and also why they are now tracking toward over $20M in annual interest expenses – nearly half their revenue on interest alone. It seems some of this expense may be in share issuance in lieu of cash interest, but whether it’s dilution or cash out the door, it feels like desperation. They’re back up near $100M in cash on the balance sheet now, but it just keeps burning…

9 Likes
6

One of the things they continue to point out is that their solution can be branded to the company licensing the product. This means that a Kia doesn’t have to pay Siri licensing premium or same with saying…with google assistant. I like their attempt to differentiate, but I agree there are many 400lb gorillas in this fight.

Yep, that is a huge red flag. I am taking some comfort from the moves they are making to reduce expenses. If this continues in right direction (already three Q’s of progress) then I think this is too much looking in rear view mirror.

Again, I agree, good point and things have to keep going in correct direction for this to be addressed. The dilution is my highest concern just because I have been burned in those situations before. But, I also see the flip side which is that companies need to spend money to grow, and as long as they get financials under control I am willing to wait a bit for profitability.

This is literally a minuscule position at moment to keep it on my radar. I have less than $1000 invested at this time. Actually, I have less invested here than I do in crypto…LOL.

Good feedback. Thanks

3 Likes
7

Totally forgot that I had already posted on this stock. I have not added to my tiny position at all, but today has a HUGE rise pre-market. Up 75% at this point, so I did a bit of digging.
It seems a other “AI” companies are posting filings that show they have taken an ownership interest in the company.

From a news brief -

“3-F filing from Nvidia in which the company disclosed a 1.73 million share stake in SoundHound”

Not sure how this is worth the huge jump in price today. I have them listed as reporting earnings first week of March. So going to be some time between this news and anything substantial.

9 Likes
8

This is leading the financial news on Yahoo right now. When NVIDIA takes a stake in something, it’s a feeding frenzy.

1 Like
9

Nvidia did not take or start a stake - it’s just the first time they had to file a 13F because their ARM holdings put them over the $100m threshold - the stake in SoundHound is from 2017, since which Nvidia has not done any buying or selling (and why would you, $3.7m is miniscule for a trillion dollar company)

This is a nothing burger

30 Likes
10

Earnings are out and they are supporting my thesis here and I want to keep this position around for some time.

I had a few thoughts about this company for the last couple weeks, and the things I figured worked out. The news about Nvidia should have been a nothing since the stake was started some time back, but the market just kept bidding this stock up.

I took over half my position off the table with the run up. This is a very small company and the stock is volatile, so I was guessing that the earnings release would tamper down the AI/NVDA hype here. After hours the price is dropping so I should have a chance to rebuild my position (only up to a tiny test position though, nothing crazy). But I put all those profits into other longer term stocks when I was rebalancing last week…so yay.

Also, this is OT, but I also bet a covered call against these earnings and it looks like I am going to keep those funds and my stock! (OT for sauldom, so do not ask for clarifications here.) I mention this because I am still treating this stock as very interesting, but very volatile and I am not worried about future buying points coming my way.

As for earnings…pretty good. The press release doesn’t match my spreadsheet I used for research, so here are some general numbers from today.

Gross margin - 77% for quarter and 75% for full year
Revenue - $17 million for year over year (yoy) of 80% (!!!) growth

They have a lot of notes about new car and restaurant customers, but they didn’t give totals, so I cannot compare with my research. I wanted to track customer growth for such a small company… :frowning: But they are gaining new customers across all facets of their TAM.

There was an acquisition (SYNQ3) that was pretty much immediately accretive to their business as it brought in more than 20 new restaurant customers.

So, they are definitely working on reducing their waste which is what I wanted to see. They are adding customers, so they are reducing concentration risk. And, they are continuing to put out new products/services that will keep customers happy.

Overall, I would say I am happy, yes, a small company can make big splashy growth numbers like 80%, but they are also working on the business things that they need to stay a going concern. I’ll def bring this back up to a small position when I see a buying point.

10 Likes
11

@dlbuffy Thank you for raising this fascinating company. I like the story here quite a bit and this company seems well positioned to capitalize with products based on generative AI. The fact that they have automative company signed for contract till 2037 is amazing.

From the last earnings call it sounds almost all positive from their business line. However, the one thing preventing me from investing is understanding their guidance as it does not seem to imply much growth.

For the current quarter they earned 17.1M (guide was 16-20M) in revenue and for the full year 2024 they are guiding towards 63-77M in revenue. Assuming they earned 17.1M each of these next four quarters that would total 68.4M of revenue or right in the middle of their range. Are they really projecting zero growth for the next four quarters, or I may be misunderstanding something here?

Additionally, they project in 2025 that revenue will be 100M, up from the 64-77M range in 2024, which also seems quite slow for the amount of business they are doing. They mentioned that this Q4 of 2023 was guided to be adjusted EBITDA positive but it came in at -3.7M, although this was explained they spent 3M+ to become a large accelerated filer for the SOX 404 regulations.

They also said in 2025 they expect to get to Adj EBITDA positive. I cannot understand how they could project this current quarter to be EBITDA positive but not 2024.

During the call they are saying “there’s a lot of good visibility” on the bookings backlog and say “we felt confident enough to give an early read of 2025” which is based on the confidence and commitment of their customers.

My gut tells me they are being overly cautious in the guidance by a large amount, although my concern is they do not say this explicity. For example, if they had said we are taking prudent and cautious guidance because of the auto market that would make sense. Or just saying we are projecting out current Q4 2023 revenue as a lower cap on our sales, that would also make sense. I’m just not confident enough here to have a good read on management if they could really be sandbagging this much on the prospects of their company.

12 Likes
12

The following examples suggest the enormous value of combining AI with language to help bridge the gap in the communication between meaning and word selection.

I was working for an American auto company engaged on a major project with a European auto company in the 80’s. Technical translation was an significant issue then. Example – conduit, pipe, hose, tubing and even straw all have a similar but different meaning in the US. A half a paragraph was required to convey the meaning of many technical terms that an experienced human mind picked up instinctively in a single word. Another, example, at the time in the US, the enclosure maintaining air flow separation between an automotive paint spray booth and the oven was frequently called a vestibule. In follow up written communications we struggled with the challenges being made by the Europeans with our design of the paint porch.

Graydrake

5 Likes
13

Thanks @dlbuffy for bringing an interesting company to the board. I found two (bright) red flags however, and was wondering how you were thinking about those:

1. Revenue growth is not expected to accelerate much and they are valued extremely richly

Their revenue and revenue growth rate of the last 3 years were:
2021: $21.2m, up 62.8%
2022: $31.1m, up 46.9%
2023: $45.9m, up 47.4%

For expectations the average average analyst estimate is:
2024: $69.5m, up 51.5%
2025: $102.8m, up 47.9%

So P/S is about 41.6 times and fwd P/S is about 27.3 times.

2. They are still hugely unprofitable, burning cash and massively diluting shareholders

In the last year net loss was 194% of revenue, in the last quarter, net loss was 105% of revenue…that’s quite something imo and makes SentinelOne’s losses pale in comparison.

They have been consistently unprofitable for the last several years (all the years I looked at back to 2019) and have funded that burn with copious amounts of share dilution.

W/A diluted shares outstanding was 11.6m end 2019 and 229.3m end 2023…that’s another record for companies I’ve looked at, I think.

So this is a tiny, hugely unprofitable, cash-burning company with a forward P/S of around 27.3 times.

There are no SaaS software companies with a valuation that high. Crowdstrike’s is the highest with a forward P/S of something like 19 or 20 times…and that is a monster company growing at 30%+ at scale with FCF margins in the stratosphere.

Why would you pay 27.3x forward sales for a small company like this that is still hugely unprofitable, burning a lot of cash, not expected by the market to really accelerate growth much and a serial share diluter?

Are you expecting sales to accelerate much more than the market, and if yes by how much? And what are your expectations about profitability going forward - are you expecting breakeven soon, and if yes how?

-wsm

18 Likes
15

Wow wsm, you are making me think sooooo much. LOL.

My first thought is that they only just went IPO in 2021/22. So the share count dilution is not as big a red flag to me because that is what you have to do to grow and get investment dollars.

Not sure of the credentials of this site, but it lays out numbers nicely:

So they show that '21 = 67 million and two years later they were at 229 million. But, the actual IPO was in 1st Qtr of '22, so at end of '22 they had 157 shares. I am looking at this as the law of small numbers? LOL. Basically, the same idea that a company with small revenue that doubles can really still mean small numbers just because it is easier to double from lower base. So, the share count is not something I am watching at moment.

An example of other share dilution I am not worrying about at moment is AXON. They went from ~54 million shares in '17 to ~75 million in 23. That is about 28% dilution in that time. Sound went from (IPO) ~162 million shares to ~229 by '23. (Slightly diff numbers based on different sites, but in range of the number for '22 IPO.) That is about 29.3% dilution. Granted it is 6yrs vs 2yrs, but by percentage it is not that far from other growth stocks in Sauldom.

(?? Please contradict me if I went the wrong way on those numbers.??)

As for profitability and revenue concerns, my thesis is the proven reduction i costs they have for a couple of quarters now. 50% growth with 23% year over year cost reduction means they only have to continue their plan to reduce expenses for a little while till the numbers flip over to operating profits. (Net losses yoy are also down by 42% which is big in my hypothesis for this company, if those numbers reverse I will probably be out.)

Right now their whole approach is getting voice response into cars, restaurant sales systems, and a few other things. The whole NVIDIA owns a stake in them part is what caused a WHOLE lot more run up than I had ever planned on. I am looking for them to continue to focus on more restaurants, expanded use in cars, and more AI voice response stuff and continue to grow in the 50% range.

None of my premise has ANYTHING to do with IOT or AI robots, but movement in that area would be amazing, but maybe it is outside of their core competency at the moment.

As for cash burning, that is the purview of all growth companies that need to focus on growth and innovation before they start having cash on hand to show ‘profits’.

Lastly, as I have stated a number of times, this is a TINY investment for me at the moment. I am even spinning options on this small holding (OT for Sauldom) because I expect the volatility and I totally expect this stock to drop back into the $4-$5 range when ANY ONE of the AI darlings hits a rough patch.

I believe they are looking to be profitable in a couple of years. This is from their earnings call transcript -

" We see the overall top line growing to within a range of $63 million to $77 million, with $70 million at the midpoint target. As we look further ahead to 2025, we believe we will cross $100 million in revenue and deliver adjusted EBITDA profitability. Our gross margins have been in the range of 70% to 80%, providing a strong indicator of our software business profile"

So this is a test position in a (basically) brand new company that is working in the world of AI by way of language interfaces. I know I have poo poo’d the capabilities of this tech on the AI robot threads, but I do see good future in things like responding to customer inquiries at the drive through, answering customer questions on in voice response systems, and easy application in cars and IOT when customers are just asking questions like, “map my route to burger king” or even “fridge, how much milk to I have?”

5 Likes
16

I guess I am not trying “to make you think LOL” but asking you for more information on two specific points given that you brought the company to the board and probably know it better than the rest of us.

Here is the latest transcript of the ER: https://investors.soundhound.com/static-files/383b1f99-cbc9-408f-afb9-d5ba354067aa

And the PR: https://investors.soundhound.com/news-releases/news-release-details/soundhound-ai-reports-record-quarter-80-q4-revenue-growth-171

And the financials: https://investors.soundhound.com/static-files/261c9dac-1126-4893-813d-b4baf123247f

I am genuinely interested in new companies being brought to the board and commend you for doing that. However given what I’ve seen of the financials of the company and the valuation I would definitely not invest.

You already stated the debt load as an issue, but this is huge imo. The debt load is more than 100% of revenue: $84.3m excluding the “current portion” and at a very high interest rate (14% I believe). This is a big no for me. Why do they have to pay such a high interest rate on their debt? Why could they not get cheaper debt? How will they likely pay it off - by issuing more equity?

And no, the dilution issue is not acceptable in my view. Using your numbers above it was 45% just last year (229/157-1). That compares to the 28% for Axon over 6 years…I don’t think those two are even worth comparing. And whereas most of the dilution in other Saas stocks we follow here are due to SBC, this company dilutes because they have to, otherwise they cannot fund the ongoing operations imo.

Their revenue profile also does not inspire me. How much of it is recurring? Given how lumpy it is, perhaps not that much? And they don’t disclose NRR.

So about the two specific questions I raised, let me try to be clearer.

The company is valued at an incredibly high multiple of revenue vs the current universe of stocks I track, implying in massive expected revenue acceleration. The analysts covering the stock have not put that into their consensus views, nor has the company in its guidance. You are currently a holder after the recent results and at the current price, and imo if you don’t have a compelling view on why revenue would accelerate, then I see no reason at all to buy. Hence my request for you to educate us on why you see revenue accelerating. If you can’t, then I personally would not touch the stock at this price.

Given that you piqued my interest, I went looking for possible reasons and found one potential answer: the very big uptick in their Cumulative Bookings Backlog, which doubled from a year ago to $661m. However, then I saw that they updated their definition of this metric this quarter (which obviously boosted it massively - see below), and I was left doubting the usefulness of the number. The old definition indicated an imminent revenue slowdown; the new definition seemingly the opposite. But wait. They basically added what they expect from future subscriptions to that number. So it is more of a pipeline number than a bookings number now. This is how they are now defining the number:

Cumulative Subscriptions & Bookings Backlog: The Company has updated this metric to incorporate its customer subscriptions activity with previously disclosed Cumulative bookings backlog. Cumulative bookings backlog takes into account the prior quarter end balance of bookings backlog plus new bookings in the current quarter minus associated revenue recognized from bookings from prior periods. Cumulative bookings backlog is derived from committed customer contracts and this definition remains the same as the previous one. Subscriptions backlog refers to potential revenue achievable for the company with current customers where the company is the leading or exclusive provider, and assuming a 4-year ramp up during which time our technologies are being implemented and assuming a successful full roll out of our technologies over a total 5-year duration. Reasonable assumptions about adoption percentages are included, with lower percentages applied to pilot and proof-of-concept customers.

→ My read is that this is the best guess total pipeline of potential revenue for the next 5 years from current customers, if all goes according to plan. However that’s still not much more than $100m per annum. So I just don’t see it.

  1. The company has stated that they will hit adjusted EBITDA in 2025 per their just released guidance:

As we look further ahead to 2025, we believe we will cross $100 million in revenue and deliver adjusted EBITDA profitability

But in Q3 2023 they also said they would hit the EBITDA positive milestone this quarter, Q4 of 2023:

we continue to believe we can achieve adjusted EBITDA positive in Q4.

And just one quarter later this is what they just said in Q4:

Excluding certain costs that were necessary to fuel our growth and transformation brings us close to the positive Q4 adjusted EBITDA target we laid out last quarter.

So they missed their EBITDA guide one quarter later just now - and adjusting the adjusted EBITDA number would have gotten them “close” to what they promised just a quarter ago…So, whereas they said they will get there in 2025, I’m doubting that given their track record. So again, given that you own the stock and brought it to the board, I thought you had a view on that.

If not, that’s the second reason I won’t be a buyer.

But that’s just me. Hope that’s useful.

-wsm.

30 Likes