Having spent the past 4 years championing digital broadcast mobile TV to the point where even we can see that it is getting a little too late to launch such services, we have always given short shrift to Telegent, the US chip company that for the past two years has been terrorizing the established order of things in mobile TV, and instead preaching that it’s fine to watch existing analog TV services on a phone.
This week Telegent has filed to go public in the US, raising $250 million, the first mobile TV chip company to get beyond the venture capital funded stage, if you don’t count Max Linear, which filed for a similar $100 million IPO earlier in the month.
Telegent is four years old and came out of stealth two years ago with its first of two generations of chip sets, which can receive and demodulate any type of global analog TV signal – NTSC, PAL or SECAM. Since then it has shipped over 50 million chips, way more than any single chip supplier in digital broadcast TV, and only in Japan has any single country shipped more devices than that (around 70 million ISDB-T 1 seg devices by year end, but this is across a handful of suppliers, of which Max Linear is just one).
Even in Japan, where we are all fond of saying that the service is a huge success for consumers, but where no-one except chip makers have ever made any money out of mobile TV, there are no chip companies that can boast that number of chips. Siano, widely thought of as being the biggest digital Mobile TV chip supplier outside of Japan, ships around 4 million to 5 million chips a year. And yet here we have a US company which is shipping 5 times that number and making a profit.
The entire mobile TV industry began when Samsung showed a phone in its native Korea picking up first an analog signal and then an ATSC signal. This concept phone never took off because of battery issues and the world set about solving the problem of how to get around this.
Cutting the story short the solution was to send a far more powerful signal, which cut out the on board power consuming amplifiers and also to send the signal as a burst, so that the on-board radio chips on a handset could process say 4 seconds of video in 0.2 of a second and then switch the radio off for the rest of the time, as that video played out, saving around 93% of the radio’s power consumption. The screen and the effort of rendering would be the only parts consuming power during playout.
This is more or less the approach taken by most of the 17 or so mobile technologies we have followed, with the exception of T-DMB and yet this is one of the two widely installed digital broadcast Mobile TV systems. Of course this approach can’t be tried in real time analog broadcasting, because it’s been set up as a continuous process.
What this produces are chips which burn (at least) 165 mW (milliwatts) of energy when all the processes are going and around 25 mW when playout only is happening, resulting in a total power consumption of 45 mW or so, which each year comes down a notch and is perhaps at 40 mW now.
Speaking this week to Diana Jovin, VP corporate marketing and business development, at Telegent, she tells us that Telegent chips burn more like 300 mW, something over five times that power. Given the power sensitive nature of the cellular industry you’d think that this would be a non-starter.
It is especially worrying when you realize that most of the handsets which have these chips in are sold in China, South East Asia, the Middle East and in Latin America ‘ places where power, and access to mobile re-charge is that much less widespread than in countries like the USA and Western Europe.
But then there are some benefits in those countries too. Most of them (China is changing) use GSM networks which require much less power for voice communication, and offer far fewer capabilities on their handsets. ‘Many of the devices that use our chips offer just voice, SMS, TV and FM,’ said Jovin, ‘and ship for just $35.’
So in many instances around the world Telegent is targeting people who have low or no access to existing TV broadcasts. But this is NOT true in China where common misconception says that it has a low incidence of TV ownership, but where it has roughly one TV per home (which means that some homes have three and some none, but most people watch TV at some time in China). This is not like the countries where Telegent has more recently opened up markets in places like Africa, the Middle East and Latin America, where some communities really do have no TV (also true in India). So the great bulk of Telegent sales (in China) are to people that watch TV anyway, and just want to watch it on the move.
It IS likely that most people in China don’t know that the rest of the world largely does not have TV on their mobiles, and buy it partly for prestige, in order to show it off ‘ a common driver in China among the newly rich – in which case battery life is perhaps not quite so vital.
‘Our chips can provide 6 hours of viewing time in most devices,’ claimed Jovin, which is just disingenuous, because that would only be true if they were fitted with extra large batteries which bulged out of the side (which they probably are).
But the neat thing about Telegent is that it is profitable on a strategy of making chips only where there is already a signal, a strategy different from every other mobile TV chip maker in the world which instead have made the chips based on promised or freshly launched services.
There those devices have commanded a premium and while Nokia continues to charge around $600 for all of its DVB-H phones (which is perhaps what’s slowing the market down, though it came out with a $250 product last week) partners of Telegent are only able to charge for low end phones.
But will this strategy last? Telegent is fond of claiming that 93% of the world’s population cannot get digital TV. It never produces the ‘other’ statistic, which is that a chunk of those people, perhaps sizeable 20% or so, cannot get ANY kind of TV whatsoever, including theirs, but that’s by the by. The truth is that Telegent is making money on a bill of materials (BOM) that has to be close to $5 per device (we get this figure from taking its past 18 months of revenue and divide it between the 50 million devices), while DVB-H and other mobile TV services began with a BOM of around $20 and has driven this to about $8 in the sorts of volumes which are still not high enough for anyone else to make money.
None of this seems to make any sense, but the common investor just looks at the numbers and pulls out his order pad. Can Telegent flourish? Can it Survive? Has it got long before others enter this, a market that all the major chip makers decided wasn’t worth the effort of entering?
Well for one thing we can be certain that eventually all TV is going digital. Across EU Europe where there are 500 million people, there is a hard cut off date of 2012. That’s over 15% of the world’s population that are not strong candidates for taking analog TV devices. Most of the rest of the world has paid lip service in saying they will switch to digital by 2015, and that includes China. If it makes this schedule then a purely analog strategy for Telegent would seem doomed.
‘We have already begun a hybrid strategy to allow the transition from analog to digital on our chips,’ said Jovin, ‘Our PCTV chip already includes reception for DVB-T.’
Now the Telegent strategy gets more confused. We can see a reason to offer a pure DVB-T chip if you are going to go to market with a story that your chips can get you TV on a chip, at power levels low enough to put in a handset (or even a Netbook), but a hybrid chip surely needs to be a hybrid technology where Telegent is strong, and that brings us back to China and Telegent’s recent incursions into South America.
In China the Government has alighted on DMB-T/H as a standard for digital TV build around a Time Domain Synchronous – Orthogonal Frequency Division Multiplexing (TDS-OFDM) architecture designed by Chinese exiles at the US company Legend. The chips produced so far are clumsy, large, heat producing and power hungry and there seems a long way to go.
There is no sign as yet that this will be anything like as successful as the digital mobile broadcast services CMMB, which has the design for mobile TV there in China. There is every chance that China is still in a state of flux over digital TV standards; it may continue development of the satellite terrestrial hybrid of DMB-T/H, or it may take another standard or it may just be 5 years late in moving to digital TV. This may bode well for Telegent or not. If CMMB eats up all the space in Netbook and PC market with its existing services, then the hybrid that Telegent would need for China is analog/CMMB, but that’s not clear yet, so Telegent has had to target a different part of the globe for its hybrid chips.
Most of South America was toying with DVB-T and one or two countries have declared for it, but since then ISDB-T services have taken hold and taken off in Brazil, and Argentina, Chile, Peru and a number of other South American countries have changed direction and backed ISDB-T. What Telegent should have done is build a hybrid analog ISDB-T chip, but since ISDB-T is primarily designed to support the shift to HD services, the processing power required is intense for full ISDB-T, so perhaps a 1 seg version would be a similarly good hybrid.
Which is perhaps why the company arrived at the decision to go with what 35 countries so far have already decided, DVB-T, even though right now there are no existing Telegent customers that could use this chip and it has to go out and convince a whole new swathe of European and South East Asians that this swing device is the right step.
The thing about DVB-T is that it is a mixed bag. It is a flexible standard with lots of different numbers of signals, operating on different power limits, with varying guard intervals, and different modulations. The receivers might work anywhere, but the quality of signal is different from one country to the next. For instance the UK network is built for rooftop antennas, and is not powerful for in-building penetration, while the German and Austrian DVB-T services have far less quality content, but are robust enough to reach car seatbacks, simply because a difference in requirement at roll out. Perhaps the 111 million homes in the US which have switched to ATSC along with Korean, Canadian and Mexican ATSC signals would give an analog ATSC hybrid a better chance of reeling in more profits for Telegent ‘ who knows?
We find the idea that Telegent is adopting a hybrid strategy rather re-assuring. If it was not adopting one, you would worry that it was confined to poorer parts of the world, where phones are cheap. But at the same time there is a quality video imperative in places like America and Europe and Korea, which defies any kind of use of analog TV. We have not seen the signal on a Telegent handset, but we know plenty of people who have, and who say that it is little better than ‘just good enough,’ and questions arise about its ability to attract customers in those markets which demand high video quality.
Jovin told us, ‘Initially our products, in places like China, were a retail proposition, but in South America we have found operators who say that it helps attract and retain clients, so they are sourcing devices directly.’
Well if you ignore Argentina, which has around 30% of people on contract, there are very few post paid subscribers in the entire of South America, so their habit is to buy a phone first and then pick an operator, and usually they buy several SIMs and mix and match. So we’re not quite sure how analog TV chips are going to help any operator there, but perhaps that part of the world is changing, and we know they do subsidize contracts in that region, even though it is a minority of business, or perhaps TV phones can be tied to a single network.
So perhaps a hybrid analog ISDB-T signal will be next, but right now we suspect that the reason the company is working on an analog DVB-T hybrid was that it was a surprise that South America went the ISDB-T route, and Telegent had expected its first digital market to be DVB-T in South America.
Pretty much wherever there is DVB-T, the region will also have 3G cellular networks and the big threat to analog TV on your phone is not a rival digital broadcast, but instead it is the increasing bandwidth that comes with successive generations of wireless broadband ‘ HSDPA and EV-DO ‘ the uptake of multicast radio services such as IMB and the gradual improvement in Wi-Fi data offload services. As 3G networks get faster, they can support more unicast streaming TV channels per base station segment, and as base station software is upgraded, these channels can be shared, and don’t have to be exclusive to a single viewer. Also where there are dense Wi-Fi clouds, customers can offload TV viewing to Wi-Fi, as AT&T does with Apple’s video services.
There is a perfect example of this in the UK, where the BBC iPlayer has around 3 million viewers watching catch up TV on a cellphone over the 3UK and Vodafone cellular networks. O2 also offers the iPlayer on some of its phones, but only allows it to work through Wi-Fi, not on its cellphone network.
So Telegent would be competing in the UK with an existing service which 3 million people already use on phones. The same thing is happening in France and by and large a DVB-T hybrid probably won’t be much good to it in the long run, except as a learning curve for further hybrids and experience in bringing down the power use.
The filing on the SEC web site makes it clear that this PCTV chip, as Jovin calls it, is only a PC chip because it uses up 500 mW of power and no-one is going to let that loose on their handsets. This is especially true in Europe where due to the 100% plus mobile penetration in much of the region, cellular operators have to fight for every inch of ARPU and fight a constant war against churn. In such markets a dead handset with no battery is lost revenue and it just can’t happen.
So realistically to invade any high margin western market, with widespread handset subsidy, Telegent has to bring that power load down to under 250 mW, at least, which it might manage in subsequent generations.
In the meantime the Netbook market and the rise of book readers which double as TV sets, will provide a strong home for purely analog TV chips, across most territories, albeit we think this is a 2 to 3 year opportunity, no more. Eventually all video will become digital video, think of how dense storage of an analog stream is by comparison, digital storage is so much better.
So what we need to see from Telegent is a way of stealing a march in all the digital markets based on the success it has had in defying logic in analog.
We are not arguing that the silicon that it produces is not exceptionally clever, it clearly is. It has a highly integrated TV-on-a-chip architecture, with an RF tuner, demodulator, decoder, video processor and all the relevant interfaces, on a single CMOS chip. And yet this is low cost, low footprint, and easy to drop into any existing design. The company claims it’s also low power, but the cellular industry right now will look at nothing above 100 mW for a single process, in any of its major handsets.
But while we find it hard to think that there are many chip companies which are planning a major invasion of Telegent’s existing markets, there are plenty of threats which could prevent it from continuing its breakneck growth and profitability as it enters other markets.
For a start there are at least three upcoming rivals which are selling SoC chips either to do the RF reception on a Netbook, and the demodulation of both digital and analog signals using the Netbook processor, or which do the entire job on a single chip, including demodulation. These chips range from $2 to $3 a piece and undercut Telegent.
The biggest of these is US company Max Linear, which has already filed earlier this month for its own IPO, aimed at raising $100 million.
Telegent had revenues of $123 million last year, and has $111 million of revenue in the first 6 months of this one. Last year it brought in a gross margin of $84 million, and in the first six months of this year it is on track to grow revenues by 80%, and has generated $39 million in net income.
At £36 million for the last nine months, Max Linear is almost half the size of Telegent, but it is well established in Japan, and making incursions into China. It returned some $23.6 million in gross margins, and in ratio terms that’s similar to Telegent.
Both companies, it has to be remembered have achieved those performances in a year when the cellular handset industry has been in retreat. The message which is being swallowed here seems to be, add cheap TV to your device and maybe you won’t suffer so much as everyone else in this recession.
But either way Telegent won’t have it all its own way once it hits the PC and Netbook. Israeli company Siano Mobile Silicon already has tied up Dell, with a generic chip for all digital TV globally, for its laptops and recently landed the CMMB chip inside the new Motorola oPhone it is launching in China. So that ties up one of the leading PC makers and a tier 1 handset vendor with one of Telegent’s digital rivals. There is no reason that Siano would not take on the analog TV market if a client paid it enough to do so, and create the ultimate hybrid, since it already supports DVB-T, ISDB-T, DVB-H, T-DMB and CMMB on its chips.
In fact cross licensing of core designs between these two companies would be most beneficial, creating the potential for each to do a global analog digital SoC hybrid, but they are so dismissive of one another that we think this is unlikely.
The major cellular chip makers like Texas, Qualcomm, Broadcom, ST Ericsson and MediaTek might also be either foe or friend to Telegent going forwards. They could license the existing analog TV designs, or come up with their own. And anyway we don’t think they will want to risk adding all that extra power consumption to their normal baseband or app processors without a guaranteed and very well established customer base.
And like Siano other specialist in digital broadcast mobile TV demodulators such as Newport media and DiBcom and Frontier Silicon will all have to consider at some point if it is worthwhile offering hybrids to help people transition from analog to the new specialist digital mobile TV format.
But will Telegent ever have the benefit of Tier 1 handset vendors supporting them instead of purely going through three or four Chinese and Asian ODMS, which make up most of its business?
‘We have already signed up with a Tier 1 handset vendor,’ said Jovin, a sentiment which was echoed in the SEC filings, but no-one would name the particular company.
Of course the Tier 1 supplier that needs a Telegent style chip most is Nokia, to support analog TV in regions where it dominates to the tune of 80% or more ‘ in Africa, the Middle East, India and other parts of Asia. But Nokia has promised it would support no mobile TV standard other than DVB-H, and although we reckon it will break that promise when a major digital broadcast market is up for grabs, it is unlikely to do so until it begins losing market share in China or India.
The most likely choice is Samsung which has proved time and again that it will support ANY technology or architecture if it will bring it market share. Motorola, while beginning to make headway in the mobile TV market, has promised to keep out of the low priced markets that nearly killed it two years ago, and LG has been pushing up-market in most of its offerings, though these days where Samsung goes, it will follow. LG is also the major author of ATSC and ATSC M/H standards and the underlying technology, so it won’t be first in the queue to undermine those standards.
All of which is perhaps why both Telegent and Max Linear in their respective IPOs at roughly $250 million and $100 million, need so much money ‘ because there is a bumpy ride ahead to marry up broadcast content with handsets and other portable devices, with many twists in the roadmap. Both have the advantage of being foundry less silicon designers that have put purely radio function into a low-cost CMOS process and that in itself suggests a promising technical lead, which has applications up and down the radio spectrum.