Ambarella (AMBA)

Question 1

DB Tech Conference 9.16.2015

Answer 1

• Recent quarter saw the ramp of a lot of new products: GPRO, Xiaomi, and two DJI drones
• Sequential growth in wearables down sequentially and YoY – “we had no intention of making a market call or sell through statements”
• “Other than the pull-forward by GPRO, it was business as usual”
• Talk about why AMBA has grown so quickly – what’s differentiated
o Video quality
o Compression efficiency – continue to see 30-40% compression efficiency vs. competitors. Can save AT&T $13 per home security camera per year because they can reduce the cost of server storage
o Power – even more important, not just battery life. More importantly is heat generation, or the lack thereof. Sensors in camera sensitive to excessive heat and leads to lower quality performance
o One example is in video – biggest issue in flying drones is stabilization, wind blowing, video can be difficult and if you don’t stabilize it, the video is unwatchable
• Are customers asking competitors to come to the market?
o Yes, all their customers want a second source for financial reasons
o QCOM came in because of the opportunity and because customers want them to have a solution to be the second source
• Revenue segments – talk about innovation in the sports segment
o Wearables, a few important things need to happen
 People demand higher frame right, higher resolution
 AMBA doing 14 nanometer chip which gives them huge benefit on the power side
o Police cameras – have been shipping police cameras in China for several quarters already, seeing a lot of activity in US and Europe. Hoping for this to develop later this year and next year (wearable camera for officers and security guards)
 This will all take time because of the back end servers, not because of the camera. You need to be able to store all of this and so the buildout of infrastructure will take time
 I think this segment and its potential is overstated – it would be a huge investment for a municipality to buildout the networking and storage to handle the high resolution 4k video that AMBA provides
• IT security cameras
o Biggest differentiators are compression efficiency (bandwidth) and high resolution (more pixels)
o What’s driving growth?
 Transition from analog to digital continues to accelerate
 Continue to pick up market share from TI design wins who left the market
 Competitive landscape is really just Hisilicon Technologies in China
• Which segment is bigger in the next two years?
o Right now on an annual basis they are about equal in $ terms
o IT security likely larger, due to home security
• Drone segment
o Started seeing this two years ago – looks like the early stage of GPRO for two reasons
 Offers consumer brand new perception of the video camera
 Both market start at the high end
 People really willing to pay $1k for a drone, and as price comes down to $499, $399, growth will continue
 Low-end side – complete integrated SoC, on the high-end side the video processor
• What happens to profitability as the price moves? – low-end drones will be lower margin due to competition
• Where are you most concerned about competition?
o IT security camera, most severe competition due to HiSilicon and Huawei, no visibility into how they fund their business, their R&D etc. “’it’s a black hole”
o Why haven’t other semis gotten involved in these markets? – doesn’t think the markets move the needle much for the bigger guys
• Auto business is all retail, camera sitting on dash cam recording video. Whole market is about 20-25mm unit a year, now seeing some OEMs and larger customers show interest
o LT goal is penetrating automotive OEM business
o Need to identify video applications in the car
 Rear and side mirrors
 Surround view (multiple cameras around the car)
 ADAS – for AMBA to get into ADAS, AMBA needs to add computer vision function to the camera
 Use their computer vision to avoid obstacles both in cars and drones
o “Moving from a video processing company to a video analytics company” – this statement was ridiculous because their competition would be MBLY who is much more advanced than they are. They said they have hopes to win OEM business but I highly doubt that considering MBLY has locked up a large part of the OEMs with a technology that’s years ahead of AMBA

Question 2

9.22.2015 Mgt. Meeting

Answer 2

Me and one other guy there who said he was new to the name. Management kept on playing up their high-growth segments such as drones, automotive, and police wearables. Home security, which is cameras sold through the likes of AT&T as part of their service, is the wildcard going forward - unclear how large that market can be. The new acquisition will eventually start competing with MBLY in ADAS (assisted autonomous driving). But this likely won’t produce revenue for another year and half. The total market size could be fairly significant based upon the applications they were talking about and AMBA may be a potential takeover target for the larger automotive chip providers

Last important piece is that Jable is a new supplier for GPRO on their newest camera, the Hero Session. This means that if GPRO is going to launch a new camera, they may use Jable in which case it would be hard to get any read from Chicony.

• Management really focused on not commenting on GPRO, several times said he will leave sports camera market comments to GPRO even when we didn’t ask specifically about GPRO or sports cameras
• IP security business (enterprise security)
o Professional security is the fastest growing category – specifically on the higher end of the market which is growing +30% a year
o How far are you on filling in TI’s market share?
 There are still some slots left to win with Hikvision and Daiwa (didn’t sound like there were a ton of opportunities so my guess is they’ve already filled in a large part of the market share)
• Vislab acquisition (small and immaterial to earnings)
o Going after OEMs is the goal, providing other video analytics for the OEMs
o Are you competing with MBLY?
 Yes and no
 Product that don’t compete with MBLY and are not currently offered – electronic mirrors (using cameras to replace side and rear-view mirrors), surround view (multiple cameras around the car – chip takes multiple camera data to show 360 degree view)
 Argued that with autonomous and semi-autonomous driving, OEMs will need to start putting a camera on the car to record any accidents to prove that it was not the fault of their autonomous systems – currently, OEMs only have an analog camera (which is archaic) and don’t even record video
 Will compete with MBLY in autonomous and semi-autonomous driving (and most likely lose I might add – though AMBA could be a good acquisition for the likes of NXPI/FSL given they already have a large amount of car content and the fact that FSL recently acquired an automotive image processing provider)
• Where do you see your business in 3 years? What will be growing the fastest?
o Law enforcement wearables – they’ve already been selling to Chinese police officers and departments over the last four quarters. Expect to be selling to the US either in Q4 of this year or Q1 of next year
 These will be sold through OEMs
o Here they again mentioned how they aren’t going to comment on the sports camera segment
• What is Chicony saying in your conversations? How large was the inventory build that caused this oversupply situation in IP security?
o Basically told me to go read their public comments
• Discuss your margins by product category
o Automotive, Drones – above corporate margins
o IP security – at/ slightly below corporate margins
o Wearables (incl. GPRO) – at corporate margins
• Newest GPRO supplier for the session was “Jable” – can’t find this company on Bloomberg and not much online but if they have a new product they are launching, Jable might be the manufacturer – this makes my earlier point on Chicony somewhat invalid if Jable is being used for new products

Question 3

AMBA Pitch 8.26.2015

Answer 3

• Upside in Action Cameras & Drones (~40% of sales combined) – AMBA makes the video processing chips for action cameras, the majority of which is GPRO. Both analyst I spoke to mentioned that GPRO sales would outperform this coming quarter. GPRO reported a 93% growth in shipments in Q2 (vs. expectations of ~64%). GPRO also raised sales guidance 9% higher than analysts’ estimates. Xiaomi camera sales should also start to contribute meaningfully to the business in this coming quarter.
o GPRO checks came in positive at big box retailers
o I asked about lead time for AMBA and they said it was 3-4 weeks – the talk is that GPRO is infilling their markets with inventory so they don’t have the same issue last year where channels were starved. I think if you combine that infill with a larger distribution network and 3-4 weeks of lead time it gives you a strong quarter for AMBA
o Drone checks are coming in positive with big box retailers having less than two weeks of inventory on hand and some stores struggling to secure steady supply
o AMBA should continue to capture all of GPROs new devices over the next year or two. Based on everything I’ve read and heard, their chip is more efficient (both from a processing and power perspective) – at least 40-50% better than any competitor
 One argument about AMBA’s stickiness is that GPRO is in the middle of developing their video software to make it easier for users to upload / edit footage. Another argument is that GPRO won’t dual source for the higher end chip because it would make the video processing quality (input) slightly different during a time where they can’t afford issues in their editing software
 The second argument is that GPRO won’t run the risk of diluting / damaging its brand with a chip that can’t do the same as AMBA – competitors are still about a year away from potentially being able to rival AMBA in the high-end camera market (the majority of GPRO sales)
• Downside in Enterprise Security (~40% of sales) – AMBA provides the video processing chips for CCTV security cameras (1080p and 4k). I think the enterprise security market is the biggest risk that no one is focused on (both analysts I spoke to didn’t include this in their top 3 list of most common investor concerns). When I continually questioned the Canaccord analyst on this segment, he agreed it is the one sort of “sleeper” risk in the equation. The analyst from Canaccord had some great insight into why this business was growing so fast. TI had exited the market two years ago and they left 30-40% market share on the table for AMBA to fill in. Combine that market share fill-in with a mid-single digit growth rate due to the conversion from analog (think cassette tape) to digital and you’ve got your answer to why that segment has been growing so quickly.
o There is a large amount of competition in 1080p which is the majority of that segments sales. High growth in 4k is not likely given all of the additional investments needed for these cameras – they produce such great quality images / recordings that you need more WIFI bandwidth and storage just to run these cameras. If I’m a gas station owner, I’m not going to have the capital nor the need to have such high quality imaging (this was confirmed through several security camera forums I read through online)
o The replacement cycle is 10-15 years on security cameras – upgrades past 1080p and 4k are not likely to occur over the next couple of years
o In Q1 2015, the enterprise security segment grew 70% YoY after growing near 100% for the previous four quarters. While Q2 2014 is a smaller $ amount to comp against, I think it’s difficult considering AMBA faces tougher competition in this segment from Huawei
o This growth rate should be lower than what analysts have modeled for the upcoming years
 Analysts have it stepping down slowly from 65% this coming quarter to 43% by end of next year – I think a more realistic scenario is ~50% stepping down to 20%
 The slowdown in this business should start to take foot over the next several quarters (difficult comps and in-fill of market share should be nearing completion)
• Other segments that could come in lower – infrastructure and automotive which together account for ~10-11% of sales
• AMBA trading around $86 right now, or 27x next 4Qs earnings (23x FY17 earnings – analysts) – the stock has traded as high as 40x over the last couple of months. I think relatively speaking this a good entry point to go long into the quarter. When GPRO announced their record quarter on 7/22, AMBA traded up to $120
• Model / Financials
o Consensus expecting sales of ~$82mm – this is a ~$11mm increase over last quarter. I think they likely come in around $83/$84 which would represent a beat of ~3%
 AMBA has beat their sales estimates every quarter by an average of ~5% - that would put sales at $86mm for the quarter. However, I think enterprise security, automotive, and infrastructure come in less than expected offset by stronger action cameras, drones, home security, and wearables
 No analyst I spoke to said that investors are worried about the enterprise security business – they are all focused on GPRO, drones, and dual sourcing. I think the first two will come in above expectations and the last concern is overblown (at least in the immediate term)
o Investors looking for a guide of close to $100mm in sales for next quarter
• The play is to go long into earnings and then sell as soon as the stock jumps. I think you begin shorting the stock the next day and throughout the next quarter. There’s too much risk during the quarter that would make the stock sell off more than +10%
o QCOM will announce their new drone chips in a month – AMBA’s chip is much more efficient (figures around 40-50%), it is unclear exactly what chips QCOM will announce. A QCOM executive said they would launch a platform for drones which includes Snapdragon that provides all flight, navigation, and photography related computing
o A report online said GPRO would announce a dual-sourcing initiative later this year – something I think would occur for their lower end cameras, not their most recent rollouts
o AMBA sells its products through two major distributors (~+90% of sales) so any issues around the supply chain would have a large effect on the stock
o From all my talks it’s highly likely GPRO doesn’t announce a new camera this year which would make Q4 a difficult comp for GPRO, though proponents argue that starved channels last year and greater distribution could surprise a lot of people. Also, if GPRO is building inventory earlier this year, it means AMBA’s sales would be front-end loaded in Q4 as AMBA sells 3-4 weeks in advance to when they get paid (total lead time from AMBA to GPRO is 6-8 weeks)
o I also think camera demand may have been pulled in considering GPRO released its new HERO4 earlier this year