Another transition year is likely for Ambarella (AMBA), but one in which it is making considerable progress with its CV (computer vision) technology. There doesn't seem to be immediate hurry to take a position, but longer-term we see potential.
Ambarella is in transition from a supplier of video processing chips, the ones that go into stuff like drones, action cameras and security cameras, to a full-blown CV (computer vision) solutions provider.
That transition has and still is taking time and a toll on the share price.
This is no wonder as the company is shrinking and operational performance has considerably worsened. The stock has suffered as a result, although perhaps not by as much as could be expected based on the above performance.
There was a brief spike in 2015 on the GoPro (GPRO) hype, but that flamed out, and its consumer business has shrunk to the mid-teens in terms of revenue contribution by the end of last year.
What's more, management argues that it will keep on declining, even if at such a small part of revenues, this is no longer very significant for investors. Needless to say though, it would be better if this segment was growing rather than shrinking.
What is growing is its automotive and security segments; although not by a whole lot, this is still mostly single-digit stuff. Revenue from the security camera market increased about 10% from a year ago quarter, led by the professional security market.
Security camera revenue represented about 60% of total revenue for the just finished fiscal 2019 (it has a bizarre fiscal year that's 11 months ahead of the calendar year) and automotive roughly 20%. The waiting is for a wider adoption of its CV solutions.
What are the company's CV solutions? It has three at the moment. From the Q4CC:
So you should consider CV2 as our high-end CVflow solution, CV22 is mainstream, and CV25 is really targeting the cost-effective solution. So for market, for example, you can think that for automotive, most of company customer will use CV2 or CV22, mainly CV22; professional security will use CV22 and CV25; and consumer IP can -- will use mainly CV25. That's just a rule of thumb... each integrating our new CVflow Deep Neural Network AI processor with our video processing into a single 10-nanometer SoC. Within seven months of sampling, we commenced CV22 mass production, and we are experiencing strong CV design activity globally.
Moving to the 10nm node retains the company's advantages in energy use and of course it already had a strong competitive position in the video processing part.
The CV market is still mostly in the engineering and design stage, not to mention the negotiating stage, but management expects revenue to start increasing in H2.
Not very pretty, with a 23% decline in revenues to $227.8M, entirely due to the consumer sector, with automotive and security growing in the single digits.
While still only generating 20% of revenue, there are considerable opportunities here, not in the least because there are so many applications. We counted five:
- Drive recorders (OEM, its biggest automotive segment, and aftermarket).
- Electronic mirrors.
- DMS; Interior drive and the cabin monitoring.
- Around View Monitor, or AVM.
- Level 2+ ADAS.
The company has specialized solutions for automotive market (Q4CC):
We also introduced our CV22AQ automotive camera assistant on chip, featuring the Ambarella's CVflow computer vision architecture for powerful Deep Neural Network processing. Target applications include front ADAS cameras, electronic mirrors with blind spot detection or BSD, interior driver and the cabin monitor cameras and Around View Monitors or AVM with parking assist.
The new SoC provides the performance necessary to exceed New Car Assessment Program or NCAP, requirements for applications such as lane keeping or Automatic Emergency Brake or AEB, intelligent headlight control and the speed assistant functions.
The quote shows the amount of automotive camera applications which the company can target. Management also announced partnering with HELLA Aglaia, a developer of intelligent video perception software for the development of next-generation ADAS based on the CV22AQ.
HELLA chose the CV22AQ because of its video processing capabilities and low energy usage, so it was nice to have that confirmed independently. Together, they are targeting multiple OEMs (which can't be revealed at this stage).
This looks like a promising partnership which demonstrated its solution running multiple neural networks simultaneously on the CV22AQ chip at CES in January.
Another partnership emerged with Smart Eye for a DMS (driver monitor system) platform, tracking driver actions and intentions.
The company demonstrated other solutions at CES like its short-range stereo cameras performing object detection while in complete darkness (using IR illumination) and its EMS (electronic mirror) solution with blind spot detection.
For the DMS (driver monitoring system), the longer-term approach is combining that with ADAS to secure things like driver attention. But there is already a customer for its stand-alone DMS solution from China.
The AVM segment is converging to a CV-enabled parking assistant solution, but Ambarella is waiting for its first design win here.
Drive recorders could become mandatory and/or given insurance incentives, so this segment is moving as well, although this too isn't imminent.
Even with the simplest processor, the CV25, security cameras can do nifty stuff like automatically recognize familiar faces approaching the front door, flag unknown persons, recognize car license plates and alert homeowners when a package is delivered.
New applications are appearing by third-party developers who use the company's CVflow platform, and many were demonstrated during CES.
Big Chinese players like Hikvision are already incorporating CV in cameras since they are a customer of Ambarella. This could be interesting this year, but management isn't aware of the split between video and CV.
The company really has an impressive position in the professional security market (Q4CC):
Our business model for professional security cameras are focused on the top 10 customers. And among those top 10 customers, we have 1 customer already start shipping in -- early this year. And also, in remaining months, we have design win and engineering activity with the majority of them except 2. For those 2 companies, we are negotiating with them to decide in terms of pricing and engagement model and -- so they can take over the project. So that 2 is undetermined yet.
Revenue in Q1 2020 is $47M +/- 3% with:
- Consumer electronics down (and it will be down for another 2-3 years).
- Automotive bottoming; expected to grow sequentially.
- Security down sequentially due to seasonality and China (the consumer part will be up).
The company expects quarterly revenue to be stable in the first half and return to growth in the second half of fiscal 2020, but there is a considerable degree of uncertainty due to China, the global macro environment, and trade/tariffs.
Despite the trade problems and some curious order patterns, the Chinese keep working with the company and granting it design wins. CV is more for 2020 (calendar year).
Management seems to confirm an analyst who argued that (fiscal) 2020 revenue might still be 10% down from that in 2019.
GAAP operating margins are downright ugly, but even gross margins have declined 10% from two years ago. Non-GAAP operating margin is still positive though, as are non-GAAP earnings ($4.5M or $0.14 per share).
Cash flow is still positive, but it's not a surprise that it hasn't escaped the consequences of the revenue decline.
The balance sheet is still very healthy though, with $359M in cash and marketable securities (up $10M from Q3), which is $11 per share, and no debt. The recent share repurchase program (of which there is still $31.9M left) has also reduced the stock count by 5.4%:
4.5x sales is not really cheap for a company that is likely to have another year of declining revenues, even if it is likely to generate cash, and buy back more shares.
Analysts expect an EPS of -$0.01 in fiscal 2020, rising to $0.58 the year after. There is a high estimate of $1.62. Needless to say, if that materializes, the story becomes a little different, although the stock still won't be cheap even then.
The company holds a lot of promise. Due to its video processing prowess, it is dominating the security camera market (especially the professional segment), but it suffers from the declining consumer market.
While that decline will continue for some time, it's now much less important for the company (revenues in the mid-teen percentage), so that decline won't hurt as much.
The jury is still out whether the company can create an equally strong position in the CV market, but the early signs (like design wins and partnerships) are really quite promising. This market takes patience though; that's more for next year.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.