If you would be so kind to embellish a little further, is there a unifying theme behind their technology?
Yes, there is, and I think that the inability of the Street to figure this out has led to a great deal of misunderstanding on the part of the market regarding what Broadcom's actually about.
As I said earlier, I'm not really a GG follower, and I haven't even finished reading the book yet. However, given the little that I know on this subject, it appears to me that Broadcom has some gorilla-like attributes. The reason for this, and the reason that the company's business is different from that of, say, Applied Micro or PMC-Sierra, stems from the issue of long-term barriers to entry.
In reality, most communications chip companies only have barriers to entry in the sense that once they get a design win for a given product made by a customer of theirs, their chip will be incorporated into each shipment of the given product; but once a customer moves to a next-generation version of this product, there's nothing forcing them to buy chips from the same company, and thus all suppliers are forced to compete on an even playing field going forward. However, this rule already doesn't apply to Broadcom with regards to a couple of the markets that it's in...and it should apply less and less as time goes on, eventually to the point where, save for perhaps the network processor and high-speed optical communications chip markets, a competitor attempting to take on Broadcom in one of its core markets may have as much difficulty as a competitor attempting to take on Microsoft in the desktop OS market.
For one example of what I mean, you have to look no further than the evolution of the digital set-top box in terms of the chips that go into it. If, 2-3 years ago, you wanted to create a digital cable set-top box with an in-built cable modem and support for wireless digital television transmissions, you'd need the following chips:
1. A transmitter
2. Two receiver chips
3. Media Access Control (MAC) chips
4. A modulator
5. A demodulator
6. An Ethernet transceiver
7. An MPEG-2 decoder
8. A graphics chip
9. Several analog chips for tuner functions
10. A standard digital television (SDTV) receiver chip
11. An HDTV receiver chip
12. A microprocessor
13. Memory chips
14. Digital Visual Interface Transmitter and Receiver Chips (for digital television displays)
Fast forward to what's possible for a company that was to embark on developing a set-top box today, and you'd see the following differences:
1. The transmitter, the receiver chips, and the MAC chips are integrated into a single, "front-end" solution.
2. The modulator, demodulator, and Ethernet transceiver are combined into an all-in-one cable modem chip.
3. The functions of the analog tuner chips can now be integrated into the digital cable modem chip.
4. The MPEG-2 decoder and the graphics chip are integrated into a single-chip subsystem (note: this wasn't possible when the DVI-5000+ was developed).
5. The SDTV and HDTV receiver chips are integrated into a single product.
6. Home networking and voice-over-IP gateway functionality have also been integrated into the cable modem chip.
7. The graphics chip can handle personal video recorder functions (i.e. Tivo, Replay TV) as well.
8. Software embedded into the cable modem chip that triples the upstream bandwidth it's capable of.
Save for #4, which has also been accomplished by ATI, all of these developments have only been made possible by Broadcom (for mass-market boxes, makes all the chips previously mentioned save for the microprocessor and memory). Also, #8 is a patented technology on the part of a company that was acquired by Broadcom, and doesn't appear to be available for license. Amazing, right? But there's more in store for the future. This includes:
1. Embedding Bluetooth and encryption processing functionality onto the cable modem/tuner chip.
2. Integrating the HDTV/SDTV receiver chip and the set-top box "front-end" chip into a single product.
3. Combining the DVI transmitter and receiver into a single chip, and eventually integrating that chip with the graphics chip.
4. Integrate satellite receiver functionality into the "front-end" chip.
5. Integrate DSL functionality into the cable modem chip.
6. (Final long-term goal) Integrate all the functions of a digital set-top box, including the microprocessor and memory, onto a single chip. Currently, the CPUs for mass-market set-top boxes are generally provided by the likes of Motorola, Intel, and AMD, who port their low-end PC CPU products into this market. Thus the CPUs used generally aren't bleeding edge, and considering that Broadcom's already been able to develop a single-chip set-top box platform (including a CPU) for the low-end market (http://www.broadcom.com/cgi-bin/pr/prps.cgi?pr_id=PR000508C), adding microprocessor integration shouldn't be too difficult to do once it's possible to combine all of the other previously mentioned functions onto a single-chip platform.
The further that Broadcom goes into its chipset integration strategy, the brilliance of their management's acquisition binge, occasionally denounced as reckless and excessive, will be understood. No company can come close to rivaling the value chain that Broadcom's created for this market, a value chain that includes over a dozen different technologies, some of which are patented. No one, not TI, not Conexant, not Intel, and definitely not a one-trick pony such as TVIA, will have a shot at providing single-chip solutions half as functional as the ones Broadcom's set to offer.
Eventually, these companies will realize what's happened, and will try to catch up, but it'll be too late, and they'll be chasing a moving target. Combine this with the fact that Broadcom will be able to leverage its value chain in a similar manner for the cable modem, DSL modem, Ethernet card, and home networking card markets (see broadcom.com for just one such example), and its gorilla-like attributes begin to show.
Eric |
