Lounging around in the Telecosm
The bad news on Conexant is that they are apparently losing the Qualcomm account for power amps to RF Micro Devices, thus lowering their share from 70 percent to around 30 percent of the CDMA power amp business that originally attracted us to the company.
The good news is that CDMA power amps are a small portion of their overall business and they are moving forward with the first onboard cable modem chip for PCs and an array of both radio frequency and baseband devices highly integrated for all the major wireless standards including GSM. They are poised to lead in Digital Subscriber Line and single chip Bluetooth transceivers, and they are well positioned for future upspectrum applications with prowess in silicon germanium, gallium arsenide and bipolar heterojunction wafer fab.
In sum, they are all over the broadband communications paradigm. Compensating in part for their current lag in share and strategic savvy behind the fabless Broadcom is possession of major fab expertise and a drastically lower valuation. 8/12/00
But the quick take is that Simon of Avanex, who is developing devices that can scale down into the enterprise, is acutely interested in securing Cisco as a customer. Cisco sells more than a hundred thousand routers a month. If it early masters the integration of PowerMux technology into routers as they are relegated to the edge of the fibersphere, Cisco's prospects improve drastically.
My guess, however, is that the communications equipment industry will repeat the pattern of the PC industry before it, where Intel and Microsoft captured the margins. In other words, communications gear will tend to break down into horizontal layers and that the value added will shift to components: chips and optics and storewidth software.
As an efficient vertical integrator, Cisco may be the Dell of the Telecosm era. But the largest winners will be the suppliers of key optical devices (Avanex, JDSUniphase et al), broadband chips (Broadcom, TI), switch fabrics/network processors (EasyChip?, MMC, Conexant), rather than edge communications routers and switches, which will become low end commodities like PCs. 8/12/00
. In all my experience, I have never seen a security class action that was anything but a shakedown. Virtually every major Silicon Valley company has been hit when their price plummeted, regardless of how exemplary. Nearly all had to settle for between $4M and $8M rather than allow their executives to submit to endless discovery and hassle, ending with a trial before clueless "peers" that consists of a "this-is-your-life" parade of disgruntled employees, ex-girl friends, professional witnesses, email garbage ferrets, and other acquisitive clowns who want you to believe that high technology companies are in it for the profits from securities scams. 8/12/00
Re:CSCO
Sure, there are many companies outside the paradigm that will do well and some within it will fail because of bad execution. The paradigm is designed to filter in a small selection of companies with extraordinary prospects because of an alignment with deep technological trends. 8/12/00
Cisco is increasingly a portfolio of network companies. In my model, it will find itself increasingly purchasing paradigm companies. Intel is also playing this role, as with Softcom Microsystems. 8/12/00
I am saying that optical demultiplexers will be indispensable in edge routers of the future. 8/12/00
re: CSCO
There is nothing wrong with being a vertical integrator as long as modular solutions are unavailable or insufficient. I am just predicting that electronic routers and switches will move to the edge and that the value added will migrate toward the chips, which Cisco neither makes nor designs, and toward the lambda processors of the fibersphere, which obviate much current Cisco software and complexity. 8/12/00
Today Cisco is ubiquitous and indispensable on the net, with products of ever greater complexity, that will be swept away by vastly simpler and more scalable optics in a lambda routed network based on lightpaths. Cisco will migrate to the edges, where the Dell analogy, though not exact, captures the essence of the new order.
8/12/00
Orthogonal Frequency Division means almost nothing. It applies, for example, to the original AMPS analog system. In this case it is enhanced by digitization and "flash" (ie fast) signal processing to allow low latency communications.
All I can say is that at very best they are a year behind Qualcomm's HDR, which has the considerable virtue of existing already, and which also uses orthogonal frequency bands (separated in time and space) and is also being upgraded to allow transactional data with lower latencies.
I suspect that like EDGE--which is already backpedaling from claims of joining mobility with multimedia--Flarion will end up as a wireless local loop provider that is either mobile or multimedia (voice and data) but not all at once.
As far as I can see, after all the political postures and machinations are over, this battle will be settled by the engineers and that the paradigm will prevail. That is, some dynamic combination of CDMA 2000 with HDR and Globalstar will become the winner. I would love to see the Ford alliance bring it to market, but as yet I have no indication that Globalstar is in the picture there (though Qualcomm did not deny that it might well fit).
8/12/00
We were discussing Cisco and other suppliers of networking equipment here. As happened in PCs, the value added will migrate from a vertical and integrated structure (IBM, DEC) to a horizontal and modular pattern (Intel, Microsoft). Collectively, however, the largest winners were companies that used PCs to change their business models, such as the firms detailed in the New Economy Watch and the dot.coms and others that exploit the Internet.
Unless uniquely efficient service providers like GBLX succeed mightily with their elasticity models (which I strongly support), noone is going to make much money in equipment or components. Crucial to the paradigm are the dynamics of the learning curve, in which more experience leads to lower prices producing wider usage and larger revenues and more customer innovations and applications in a positive sum spiral. Smart investors will try to find big winners throughout this double helix of capitalist creativity. 8/12/00
Although I agree with Russo on SONET and meshes, no one can afford to read packets anywhere but on the edge, and with lambda networks no one will have to read them. 8/12/00
No Cisco will not go away. It will bloat into a huge smorgasbord of networking investments. In time the value of that portfolio will be dominated by paradigm companies.
8/12/00
Re:AVCI
I doubt they have any enduring edge. Most of their targeted slots will soon be filled with optic cross connects. Avici will have to move to the edge where their architecture is too complex and too bit rate and protocol dependent to be scalable.8/12/00
I also will readily confess to some embarrassment at some of the literature that goes out under my name from Forbes, and I am in the process of having some of the more extravagant language muted. However it is hard enough to write this report; I do not have the time or the talent to do the marketing. With more than 70K subscriptions, Forbes is apparently doing it well.
Whatever claims were made in the past, incredibly enough, they were mostly fulfilled. Last year the list contained six out of the ten best performing S&P 500 companies, including one and two, and four out of the top eight NASDAQ 100 companies, including the top two. My two top choices, Qualcomm and Uniphase, have risen some 25 fold between them since I began celebrating them in the strongest possible language ("Intel of the Telecosm etc.") This year we launched the Digital Power Report and it has produced a list which is up 175 percent or something this year in the face of a sluggish NASDAQ. The year 2000 still has five months to go. Your four investments--Avanex, Storage Networks, JDSU, and BRCM--are among the best companies of 2000. So are Corning, Nortel, and Sun. All are likely to suffer slumps at some point. I hope we have given you and our other subscribers who hold any of these companies confidence to maintain or even expand their positions during the inevitable downturns. Globalstar/Loral, GBLX, Qualcomm, and Mirror Image are currently out of favor despite a general improvement in their prospects. Also in the doldrums are companies I have strongly cited on the board: WFII (expanding its rare prowess in deployment and maintenance of 3G wireless) and Sonic Innovations (marketing Carver Mead's unique hearing aid breakthrough). These companies still represent long term value.
In general, I believe that there remains immense opportunity in the paradigm, which requires the transformation of the entire global communications infrastructure at a cost of trillions of dollars over the next five to ten years. I will point in the right direction most of the time. But I am not going to make an inevitably futile effort to do the impossible, which is timing the stock market or judging the risk profiles appropriate for different investors. 8/13/00
EasyChip is a startup with an extremely innovative (highly parallel, super scalar) network processor design that runs at 10 gigabits per second "wirespeed" for all seven network layers. It is programmable for any protocol. I cited it as the kind of device that can render routers a commodity, as the value migrates from the integration (Cisco) to the chip vendor (EasyChip, perhaps). 8/14/00
Without any special knowledge, I suspect Wave chose ARM because it offers a very compact low power microprocessor core with many development tools, that can be integrated readily on smart cards and other mobile devices with Wave's Embassy encryption engine. 8/22/00
As an outsource for complete wireless network management--from concept to construction, and even maintenance and operation--WFII is an entirely different kind of company from Leap or Nextel, which are service providers. Both Leap and Nextel, I believe, are customers of WFI. The customers of Leap and Nextel are end users. 8/23/00
I don't know about Novalux, but IPG Photonics supplies lasers some 20 times more powerful than most prevailing pumps, which are around 200 milliwatts to 800 milliwatts. 8/23/00
I have not investigated Cube for months, sorry. I should do it. As for Wave, I am flummoxed as usual by the insider information laws. I am on the board of Wave and will be attending a meeting on Friday of this week. The company has made impressive progress this year (chiefly through the Enable acquisition and new financing), but it remains risky until one of its products generates substantial revenues. 8/23/00
Terayon is gaining share without winning the standards battle. Standards like history are written by the winners. But I do not see any large movement coming on that front during the course of this year. Terayon's technology retains a significant edge, recently confirmed in Hong Kong. I hope the strategic picture will become clearer at Telecosm, where Zaki Rakib will appear with Henry Nicholas of Broadcom. 8/23/00
re:NMTC
I have no direct knowledge of the company, but for many years it was assumed that optical lithography (projecting the chip design on the chip and miniaturizing it through a reversed microscopic lens array) could not work at dimensions smaller than the wavelength of the light. In recent years, by manipulating the phase of the light, it has become possible to inscribe .13 micron geometries with infrared light beams .183 microns (1800 nanonmeters) wide. The product described in your note would allow a further step forward to .09 micron dimensions. Every photolithography company has a program to do it. The leaders, last time I looked, were still Nikon, ASM, Canon, and perhaps Ultratech. I don't know the role of Numerical with these firms 8/23/00
Re: MRVC
Is this a company or a portfolio? I see no way to evaluate such a magpie's nest, except that I do have a position in Kleiner Perkins latest fund, which invested in Zapphire I believe and perhaps some others. But MRV is so dispersed in its holdings that it is an unlikely choice for the list. 8/26/00
Spectrian was once on the list for its innovative power amps which were used in CDMA phones. When the company lost key contracts, we took them off. As I recall it was an execution problem. But I have not inquired recently about their progress. Any news I should know about?8/31/00
. A couple years ago back in the days when I used to invest occasionally in companies after I put them on the list, I invested in Intentia as an aggressive Java play. The company is creative and I am glad to hear that the Java product is actually being introduced at last. The price is still near what it was when it was listed. 9/2/00
Denny's Law of Queues is consistent with Say's Law of markets: Supply Creates Its Own Demand. You understand that law and you never pay any more attention to the "consumer demand" estimates which allegedly comprise 65 percent of GDP. Demand does not create supply. Yet most of economics is still based on an erroneous demand model, consisting of consumer demand, investment demand, government demand and net export demand, all conceived as forms of spending. In fact, spending is irrelevant. What matters is productive services, innovation, and creativity. 9/2/00
Terayon is a fascinating company with excellent engineers, proprietary technology, and an evolving broadband systems strategy that I do not fully grasp. I hope to have further reports after Telecosm. 9/2/00
Having finished the September letter and perused the first hard copy of Telecosm, I have spent much of this afternoon in a mostly futile effort to get on the board. It appears that once again we have fallen seriously short of computer power, or Bob Metcalfe was right about the coming Internet crash, or in her famous flight from French taxes, Laetitia Casta has appeared again on a secret competitive page. I look forward to salvation by Bandwidth Angel Ellen Hancock of Exodus sometime next week, when our new servers are turned on. As for your question, the reason we don't post in the evening is that companies put on the list will be immediately publicized on CNBC or Yahoo and our small army of subscribers will be swamped by the large armies of Yahoo and CNBC. Although our choices are not made with price and timing in mind--their superiority as investments should play out over a period of many years--many subscribers understandably want at least a fair chance of buying before the herds stampede into our listed company. 9/3/00
Wave's Embassy is not chiefly a co-processor; it most
comfortably resides on the processor silicon itself, as in
the case of the AMD microprocessor. Steven Sprague's
ambitions are ample to suit your appetite for onchip
ubiquity. Although this is highly contested real estate, to
say the least, most processors on the edge of the network
will have to incorporate a security and ecommerce component. 9/3/00
In late breaking news, Ming Louie, previously of Globalstar, has just been appointed head of Qualcomm China, and he is headed for Peking full of assurance that he can turn around the CDMA Bear. This game isn't over yet.9/6/00
Off the noggin nib I would say that government anti-trust vandals have foiled Worldcom's original horizontal strategy by forcing spinout of MCI Internet and barring purchase of Sprint. Prevented from using the Internet to attack the telco establishment worldwide, the company has decided to diversify vertically instead, purchasing Intermedia-Digex web hosting capabilities which are probably compatible with UUNet. Dave Dortman tells me that Digex has superb marketing resources but inferior technology. Perhaps, the marketers will do well selling UUNet capacity. I think the old strategy was better, but as usual government is the dog's best friend, defending monopoly telcos against all comers. 9/6/00
Exodus provides storewidth from hosting centers situated optimally for the lightspeed delays that will account for most latency as the fibersphere is deployed. Exodus centers will not be obsoleted by all optical networks, while network accelerators focused on outwitting the snags in the current internet will no longer contribute. 9/6/00
My sole interest in LanOptics is as the owner of a dominant share of an Israeli company called EZchip that has conceived a very elegant network processor which was described at the May InterOp in Las Vegas. I discussed the design in Mid-August in the Forum (as part of a long and interesting exchange on the future of Cisco). If the company can gain access to a suitable foundry and deliver this device in a timely fashion, and if its application layer can in fact perform the claimed accounting, load balancing, URL switching, and content addressable memory tasks without requiring onerous changes in existing router software, and if none of the many other network processor companies and aspirants--from Motorola and Intel to IBM and Vitesse to Conexant and MMC Networks (to be acquired by AMCC)--fail to excel this design, EZchip could become an important force. 9/6/00
Novell does not chiefly use algorithms to outwit the existing network. It offers caches and other devices to accelerate delivery of material to that network. Thus Novell's directories and caches will work better as the fibersphere is deployed, while accelerators based on choosing better paths through the net will decline in importance. 9/6/00
I am not negative about EZchip. I think it is one of the most impressive chip designs I have encountered. But it will not see silicon until next summer and is in an extremely competitive market. I know very little about LANOptics except that it is the Israeli company that owns most of EZchip. 9/6/00
I obviously cannot tell you ahead of time what company I am listing, if any. Those who pay close attention to what
happens on the board will usually have a better view of my
thinking during the course of the month than those who don't.
I think I like what Corvis is trying to do. But the company
has been very secretive and so far I believe that Avanex has
a better solution for long haul communications. However,
both companies will be represented at the Telecosm
conference next week (to be broadcast on the web) and
perhaps I will obtain new insight there. Charles Burger, our
optics analyst, has just returned from the Denver optical
conference and also has alot of ideas, which he will be
expressing on Wings of Light. 9/7/00
I am very sorry that my clarifications ended up confusing people. In the face of what I sensed was a massive swell of assurance that this was a new Avanex, I tried to convey that although I am enthusiastic about the technology, it is still just a schematic and beset by many competitors. When "Doc" wrote that I was negative, I answered strongly that I was positive. I don't think anyone should have sold a company with a $100 million market cap on the basis of this cautionary posting. I was trying to create as realistic a sense of the risk profile of this small company as I could. Ultimately what happens in the first day is trivial compared to determining the quality of the technology.
EZchip is so paradigmatic--like TeraBeam so precisely fulfilling the Telecosmic model--that I find myself reifying a system that is still to be created. In both these cases, execution is going to be crucial. 9/7/00
I am sorry to have misled anyone. I wanted people to invest knowledgeably rather than assume that this set of PowerPoints heralded a new Broadcom.
There really isn't much more to say about the company. I heard the InterOp presentation, did an intensive study of network processor technology, and did extensive interviews with Eli Fruchter. They have what seems to me decisively the best conceived and most paradigmatic of all the leading network processor proposals. Critical is the memory architecture, on chip storewidth based on embedded DRAM, which I have heralded in at least two GTRs. Network processors are going to be widespread. Now the company has to execute. I tried to put it in a larger perspective in the report. 9/7/00
I don't know how to convey strongly enough that this is a technology letter. I was reacting to what I saw as a stampede into a stock by people who did not understand that EZchip technology was still just a schematic. I was not negative. I said this company, with a $100 million market cap, has the best network processor design. Anyone who sold at that point was not investing for the long term. This board is not about Gilder spike profits on the first day.
I suppose I should once again give a disclosure, since the usual charges are being made. Although long in the past I acquired shares in companies that later joined the list (Qualcomm) and later in companies that I had put on the list (Global Crossing), I now routinely turn down lucrative offers to invest in GTR companies. My holdings, all originating between two and ten years ago, include only Qualcomm, JDSUniphase, Atmel, Global Crossing, and Conexant among listed firms. I also am on the board of Wave Systems and have holdings in various private companies, venture firms. My pension is handled by Velocity Capital and I have only the vaguest knowledge of what companies they choose. Contrary to many loud public claims, I own absolutely no part of TeraBeam. 9/7/00
Ming was at Globalstar from the outset, focused on technical management. Now he is moving to Qualcomm, which is an affiliated company. He is merely performing various assigments as needed within the CDMA camp. 9/7/00
George Soros turns out to own ten percent of LANOptics.
9/7/00
I realize today that my first post (in response to someone who imagined that LANOptics firewall technology was telecosmic) seemed more negative than I intended and many did not see my correction a few minutes later in response to "Doc". (One more positive sentence got unaccountably dropped in the course of the posting--there should be an edit button). I was trying to tell the truth that this company, though promising and paradigmatic, is not a sure thing. However, I do not understand why anyone would sell shares of a $100 million market cap company with "the best chip design I have seen in years" or some such, because I pointed out the risk factors. 9/7/00
Fiber, though, where feasible offers bandwidth many orders of magnitude greater. Within the next two years or so, MFNX will have fiber links in reach of most of the 20 percent of US urban buildings that provide 80 percent of telecommunications traffic. 9/7/00
I do not do financial due diligence on a company before I list it. I choose a technology--in this case EZchip--and then my staff (or in this case folks on the board) find an investment vehicle that owns it. I still have no idea what LANOptics does beyond the general categories of firewall and Token Ring just as I did not check out any of the other companies in Xcelera before I listed Mirror Image. We are a technology letter and we believe the best paradigmatic technologies ultimately prevail. 9/8/00
Nimble noggin, my answer would vary with my enthusiasm of the day, but certainly Avanex would be highest on the list at the moment. Others would include GlobalStar--with an acute short-term risk followed by a huge long term updraft; Global Crossing which is a long term play without the same meteoric prospects but with a relatively safe fiver; and swinging hard for the stands, EZchip and Wave (disclosure, I am on the board at Wave and have several hundred K worth of GBLX). 9/9/00
But financial analysis--as I recall--would have barred listing Avanex, Broadcom, Globalstar, Global Crossing, Mirror Image, Atmel, Procom, TI, Qualcomm and EZchip--all with financial negative net worth at the time, with big debts and elusive assets--while affirming Motorola, Agilent, Lucent and all the other usual suspects. What I try to do is much more important than the letters that "just tell you what stocks to pick." I try to give my readers the confidence to buy stocks when they are low. Anyone who invests on the basis of rumors or financials or technical analysis will sell when a stock collapses. If you know the intrinsic technology is solid, you can buy at the bottom and sell at the top, rather than the other way around. That is crucial to successful investment. As Christensen shows--with many companies reaching their financial peak just before collapse--financials tell you what already happened and are almost useless for assessing future prospects. 9/9/00
In my judgement, apart from Avanex, which had no revenues or customers when I picked it, the two best technology picks I have made since the initial Qualcomm/Uniphase duo are Terabeam (admittedly private but with a forty fold appreciation in the weeks after publication); Mirror Image (with a general acceptance in the industry after the pick for ascendant technology); and Globalstar, massively in debt, like GBLX.
With a sensational design and no silicon, EZchip is a perilous play (that's why I urged caution). But its valuation is now one hundredth most of its key rivals in the hottest sector of the chip business. By the time the kind of appraisal that you seek could be made, the company would probably have been purchased by Intel or Broadcom and the management, whatever it is, fully replaced.
I say this with full and painful awareness that many subscribers got burned by timing errors and mishaps in several of these companies. But knowing the intrinsic values, other subscribers could buy these firms on the dips and far outperform most other investors who study the literally thousands of letters that focus on the rearview mirror of financials.
My view is that management and financials follow good technology rather than lead it. If you don't like Amadon, for example, he already made his contribution and he is out, but the technology remains. 9/9/00
In my view, there no longer exists a "semiconductor industry" which undergoes the usual cycles. Each of those chip companies faces an entirely different set of intersecting cycles. The companies that focus massively on PC sockets, such as Intel, AMD, and Micron, will follow the cycle of the PC industry with the usual lags and accelerators. National and TI have so large a spectrum of customers that they respond to the overall cycles of the global optoelectronic economy. But they also command specialized capabilities that yield faster growth and dominate valuation (mixed signal analog) and communications DSPs. The pure play communications companies such as JDSUniphase, AMCC, and Broadcom each have their own customer set which is almost totally detached from the "cycle." So does Qualcomm spinco, now chiefly a chip company, and Conexant, which are in the volatile but explosive wireless arena. Atmel is still engaged in a successful transition from memories to communications that is complicated chiefly by the increasing use of silicon memory in communications. 9/9/00
*********************
The newsletter seems to be a struggle to try to pick the home run and the top S&P stocks. Soros news is interesting. Alot to digest since someone was in a posting mood last month.
Jack |
