Nimish SawantMar 02, 2016 18:25:22 IST
MediaTek has been a popular name when it comes to mobile phone chipsets. Its latest generation Helio X20 chipset has a ten-core CPU which MediaTek claims comes with a triple cluster configuration thanks to the division CPU cores into a single dual-core cluster and two quad-core clusters that make up 10 cores.
The other chipset line being the P10 series is targetted at more mid-range devices. While last generation we saw a lot of flagship devices sport Qualcomm's chipset, MediaTek promises that 2016 will be different. We have already seen the launch of Zopo Speed 7 at MWC 2016 with Helio X20 chipset, Gionee Elife S8 with Helio P10 chipsets inside them. MediaTek claims that there will be many more products with Helio chipsets inside them.
While Qualcomm has the lead when it comes to the LTE chipset market, MediaTek has been growing in that space. We had a free flowing conversation with Finbarr Moynihan, general manager of international corportate sales for MediaTek, as well as Kuldeep Malik who heads the India sales division of MediaTek, to get to know MediaTek's future plans and how it plans to ensure that more MediaTek chipsets are seen on flagship handsets this year. Moynihan was attending the 'Make in India' fair which took place in Mumbai couple of weeks ago and we interviewed him then.
What is the agenda for the visit this time?
Finbarr: So we have Kuldeep who looks after India and I look after the MediaTek international sales. India is one of the more exciting and growing market. The 'Make in India' initiative and this transformation to 4G/LTE is exciting. So I have been coming to India on a fairly regular basis and this time in particular was to coincide with the Make in India week. Just to try to understand what is happening on the ground and what our customers are doing. Already a lot of smartphone manufacturing has moved here, so we are seeing what we should do. There aren’t any specific announcements as such.
Since smartphone making has already moved here, how long do you think before fabrication plants can be set up here?
We are a fabless company and get our chips are made via TSMC, GlobalFoundries, UMC and others. But if you look at how the fab plants have developed in China and where the Indian industry is today, India is just moving into the first phase of handset manufacturing, so there are many more steps to go before there can be fabrication plants here. Right now you have handset assembling, the next step would be making and assembling PCBs, followed by component manufacturing as well - including camera, battery, display and so on. Then comes the part about actual R&D of various companies happening here. So I think the semi-conductor piece of the puzzle is further down the line. It would take at least 5 years before that happens.
Do you think it also has to do with the stringent requirements of having fab plants - clean environments, huge investments, etc.?
Finbarr: No, I don’t think that is really an issue. My understanding is that one of the main objectives of the 'Make in India' program is employment. If you take any modern state of the art fabrication plant, it does not really employ a lot of people. It is quite automated for a lot of processes. The less people you have in a fab plant the better it is to keep the environments clean. So it’s not really about more employment per se, but about the evolution of the industry. Also say somewhere in the near future, if India is able to fulfill the objectives of churning out say 500 million smartphones by 2020, then comes the part of assembling, packaging and testing of the semi-conductors which will happen first. And then the fabrication plants come into the picture.
Kuldeep: The 'Make In India' plan is to have around 28 million jobs by 2020, and have a zero net import by then. But I think if you look at the way these industries have developed in China, then it has to grow in phases. What we are seeing at the moment is just the first step. Component manufacturers will only set up shop over here if there is someone buying from them, so we still have to get to that level. The best example is from the Auto industry. When Maruti started making cars in India, they also started making ancillary units around their plants. Slowly the private players got into the fray as well. The entire ecosystem developed over a period os 15-20 years.
Finbarr: Also if you look at China, it is really only now that China is starting to make big investments in the semi-conductor industry. Just like you have the momentum around ‘Make in India’ and ‘Digital India’ programs here in India, the Chinese government is also driving a specific set of programs to give a boost to the semi-conductor fab, semi-conductor design, semi-conductor packaging industries. So they are beginning to move into those areas, so it will definitely take some time for India to get there. There is definitely scope for India as it has a demand in the local market and it also has the potential to export products made in India to the Middle-East and African markets.
The number of chips you have sold worldwide for mobile phones so far? Your revenues for the year
Kuldeep: We sold about 400 million smartphone chipsets and over 150 million of those were LTE supporting ones. In 2014 that number was around 30 million LTE chipsets, so you can notice the growth. In 2016, although we haven’t made any prediction numbers wise, I think there will be a higher percentage in LTE chipsets shipments.
Would it be safe to say that going forward, MediaTek will only concentrate on making 4G chipsets across its product portfolio, or will 3G/2G still be given equal care.
Kuldeep: If you look at our roadmap, we have pretty much made most of our chipsets LTE compatible for mobile phones. But we will still have 3G for sure and that is going to stay in India for at least 2-3 years. So we will have 2G/3G platforms as well, but majority of the focus and R&D will be towards 4G. If you look at 4G, India sold around 20 million 4G enabled handsets towards the last quarter and that momentum will continue with the entry of new 4G service providers in the market. So we are trying to place ourselves there so that there are no surprises. Also a lot of the service providers have invested a lot of money in the spectrum, so 3G services will be mainstream for a long while.
Finbarr: We had announced one 3G platform in mid-2015 and have announced around six new LTE platforms, so you can see where the trend is headed. I think we have also been instrumental in getting down LTE to even the entry level handsets and we will continue doing that. But last year we still shipped around 300 million feature phones - so that is still a huge market. I believe we will still be shipping large volumes of 3G handsets for at least 2-3 years.
Last year, majority of the flagship smartphones sported Qualcomm chipsets - despite the bad rep that Snapdragon 810 got from the industry and critics alike. But we didn’t see many Helio X10 / X20 product launches. Why was that?
Finbarr: We are making a lot of progress, like for instance we announced the Helio X10 at Barcelona last year and a lot of products were seen with that phone. This was followed by the P10 and later by the X20. We apply Helio only to the highest end spectrum of devices. So Helio X is the extreme and P is one step below it. We are seeing a lot of design wins with the Helio X10 from companies such as HTC, Xiaomi, Vivo, Gionee and so on. And now with P10 starting to ship which will be followed by the X20 couple of months later, you will see a lot more brands using Helio chipsets, so we will be competing in the high end segment.
So Helio will not be seen on entry level or mid-range devices?
Finbarr: For us Helio is a sub brand, within the MediaTek family, which only represents the higher end segment. I think that differentiation between chipsets based on the modem are over. The real test going forward will be around the user experience. That’s where a lot of our core-capabilities coming from our home-business such as TVs, DVDs, multimedia, video, camera, display - all of those kind of features become more important. That’s why Helio is only at the high end as that is where we have the right kind of price proposition that can deliver the best multimedia experience. So most of the companies are still releasing Helio X10 based products, but you will start to see X20 based products as well. Also Helio P10 based products have started to go into production as well, so we will be seeing a lot on that front as well.
But Helio X20 was announced around Computex 2015, but there aren’t really any products with that chipset selling the in the market, as compared to say a Snapdragon 810.
Finbarr: So we announced the X20 in June and it only went into production towards the end of the year. Now we are just waiting for the first customers to launch the products. We announced it early as we have our announcements coinciding with major technology events, so Computex was the last major platform to showcase our Helio X20 roadmap.
So one of the reasons, Qualcomm has an edge over MediaTek is beause of their modem technology. With the Helio X20, do you think you have levelled the playing field in that department?
Finbarr: To be fair here I think the Qualcomm products still have a lead, but the question is where is that important. There were really two things: one was the LTE modem technology where we have added carrier aggregation and the second thing was adding CDMA capability. So in the past when we didn’t have CDMA support, our phones were not supported by China Telecom in China and Verizon or Sprint in the US. So for the phone brands that was a big problem, because if they wanted to have a global launch of a particular device then they had to go with a chipset which supported multiple regions. If we did not have LTE or CDMA back then, then we would definitely not be a first choice for some brands. We have since closed that gap significantly in terms of delivering the modem capabilities that’s required today for the vast majority of networks globally. The benchmark for me is, if you look at the iPhone 6 today, it has carrier aggregation and LTE Cat 6. So if it is good enough for the iPhone 6, then it is good enough for us.
You have a deca-core CPU as an offering in the Helio X20 lineup. Is it really necessary to have more cores on mobile CPUs? Are there really many apps that actually stress all the 10 cores? Wouldn’t more cores lead to more heating? Could you elaborate on the power/heat management in Helio X20?
Kuldeep: When we were designing this SoC we realised that there are many simple processes such as using a calculator, playing some audio, etc., which does not require that much CPU/GPU power and can be done more efficiently. With our tri-cluster platform we redefined the big.LITTLE architecture. So between the big and LITTLE we have put in a medium, power usage wise. With this implementation we realised that we had around 36% efficiency. So there is a lot of scope for optimisation, which can lead to using a slim battery and so on.
Finbarr: We don’t imagine any application will use all 10 cores. So instead of having two clusters in the big.LITTLE architecture, we give you three clusters which gives a more granular approach to using apps based on their load. It is still an ARM processor, but there is always a trade-off with CPUs - speed and power. The faster the processor, the more power consuming it will be. So if you make a low power cluster, you can just optimise it for low power. High end cluster you can optimise for the extreme requirements. But there is still a gap between the low power and high power clusters, hence we decided to use a tri cluster design to make it even more efficient. So in some cases the cores will be the same but we decide the frequency at which they will operate. So you have more granular steps across the use cases. It is our scheduling hardware and software that sits on top of the cores, between the OS and the applications, which does the scheduling. We use our CorePilot technology in the Helio X20 as well.
So MediaTek also makes chipsets for TVs, so could you tell us how big that market is, because in India MediaTek is only associated with mobile phone chipsets.
Finbarr: If you take MediaTek as a company it is about two-thirds mobile and one-thirds home. Home includes not just TVs and blu-ray players but also some gateway platforms. Now, last year we acquired another Taiwanese company called MStar, which also has a TV business. So even though it is a 100 per cent subsidiary of MediaTek we run it as a separate company. So we work as separate companies. And if you include that, it is close to 50-50 for mobile and home. In the TV segments we provide chipsets for Sony and Sharp’s Android TVs, Panasonic’s FireOS TV, we have supplied to LG in the past, Vizio in the US, Philips, TCL and HiSense in China. Then if you go to MStar, they supply to Samsung. So between the two of us, we have pretty much all the major TV brands covered. But as far as India is concerned, we are concentrating only on mobile phones, because if you look at numbers last year - 280 million mobile phones against 12 million TV sales - it pales in comparison.
Since you have been operating in the Indian and Chinese market, what are the challenges between the two markets
Finbarr: China has undergone a fabulous transition in the last two years - going from completely 3G to completely 4G now. 3G smartphones do not sell in China anymore, it’s just 4G phones. This is obviously driven by the operators, the government, the carriers and so on. So that is quite exciting. In the last couple of years it has also become a high tier market, so phones with very high specs sell a lot there. Consumer awareness and consumer demand for those kind of high end specs is driving Helio level quality. Infact all of the entry level smartphones that are produced in China are exported. 250 million 3G phones that were sold with MediaTek chipsets last year - almost 95% of those were manufactured in China and probably all of those were exported. So that’s the position in China. The challenge is that the market in China is slowing down, there isn’t rapid growth. Around 450 million units are sold annually. But the curve is flattening and you are getting into the replacement market and so on.
India, on the other hand, has around 250-280 million buyers annually. Most of the sales happen in the entry level and feature phone segment. So to put it simply there is a lot of scope for mobile unit sales numbers to grow. If India’s GDP keeps growing at the rate that it is growing, if wealth continues to be generated in the country, I think there is a lot of opportunity for mobile business. And I don’t think we know what kind of design innovations we will see in the next few years.
Kuldeep: This year’s forecast is I think around 133 million will be smartphones, 67 million will be LTE based handsets. There will be challenges from local brands. Today around 51% share is still held by the top 10 local players/brands. But they need to up their game. You can’t just import phones from China, rebrand it and sell them here. There has to be investments in R&D departments of these local brands. Companies have to go beyond the pricing and look at improving overall user experience. Differentiation is key.
You have the LinkIt platform for IoT products and at Computex last year, I saw a lot of implementations across different product segments such as bike helmets, bicycles, clothing and so on. Are there any specific requirements that developers have to follow? How is the work on the platform going?
Finbarr: LinkIt is part of our overall approach to IoT. We formed a new business unit (BU) focussed on just the IoT last year and it is a very small unit as compared to our other divisions. There are four produce areas: Wearables, Home Automation, Machine to Machine (M2M) and GPS. We sell a lot of stuff in India for energy meters. So this BU is focussed on developing products for these four categories. Now we recognise that IoT is a whole different landscape as opposed to the mobile landscape. There are a finite number of customers who drive the volumes in these fields. But in the IoT space, it is a very nascent stage and it is a very broad field. So we are adapting our approach to the market. The MediaTek Labs is one approach by reaching out to developers. It is more about seeding these projects. Then the business side deals with how to market the product if has the potential to become successful.
In addition to Qualcomm and Intel, you have Samsung doing very well with its Exynos chipsets and even Huawei has started making its own APs. Do you view that as a threat to MediaTek’s market share?
Finbarr: See there is always going to be competition, I mean we are not the only supplier in the market. If you go back in time in the mobile industry, Nokia, Motorola, Siemens, Ericsson had their own chipset businesses. That’s when MediaTek had also got started with mobiles and grew from there. Then there was a time that no one was doing things internally. Till Apple came along and developed their own custom chipset, Samsung followed with its Exynos line and so on. But there is still a very sizeable market out there. Also if you look beyond just the mobile market, there are a lot of other areas where you can leverage the potential of your chipsets. If you look at the broad Android platform, the mobile platform, it will certainly expand beyond smartphones. The best example would be the auto segment. A lot of embedded products such as payment terminals can leverage Android. So although the premium market is still quite vertically integrated, there is still a sizeable market from companies out of India and China which give us opportunities to grow. As long as we are developing the right products with the right differentiating features, there will always be a market for us.