Friday, May 8, 2015

High Tech "State of the Union", Q1 2015 - Part 3


Look not mournfully into the Past; it comes not back again. Wisely improve the Present; it is thine. Go forth to meet the shadowy Future without fear and with a manly heart. - Henry Wadsworth Longfellow

A full five centuries before Columbus, Leif Ericson strode upon the shores of the new world. Following rumors of lands sited on the western horizon by sailors blown off course from Greenland, the Vikings discovered Labrador and Newfoundland in eastern Canada and, according to the sagas, set up several colonies all the way down to the Saint Lawrence estuary that traded in timber and grapes with Viking settlements in Greenland, Iceland and beyond.

To do what these men did took an enormous capacity for optimism and complete disregard for personal safety. I have myself crossed the North Atlantic in winter, but in a modern ocean liner; for someone to do the same in a Viking longboat strikes me in comparison as being completely lunatic. 

Consider also the times in which Leif and his crew performed this recklessly daring feat. Europe was still deep in the throes of the Dark Ages. Charlemagne had been crowned by the pope just two centuries before as the first Emperor in Western Europe since the collapse of the Western Roman Imperium. Even in those wretchedly trying times, there were men possessed of stunning bravery and vision like Leif and Charlemagne who sought to explore new frontiers and build the foundations of peace and prosperity for future generations.

The Carolingians

And how are Europe's modern day explorers and adventurers on the frontiers of Technology doing as they probe far horizons of their own? The numbers are in for Q1; what they tell us is quite intriguing.

As we've seen in previous "State of the Union" editorials, this trio of firms mirror each other in terms of technology and market focus - analog, mixed signal, specialized discrete components and microcontrollers all targeted at industrial, automotive and IoT applications. All three are also very much vertically integrated like the semiconductor firms of old, with the majority of their foundries and assembly & test facilities wholly owned and on European soil. Yet differences have emerged between them - some subtle, others much less so.

Infineon completed its merger with International Rectifier in January. The company's second serial quarterly jump in revenues is due in great part to the final integration of the acquisition. The merger strengthens Infineon's positioning in automotive and industrial sectors for power management and motor control. It also gives the company a very significant footprint in the North American market - the world's largest IC market - as IR's manufacturing back end is almost all located in the USA.

Though the latest financial results seem to have reversed the relative positions of the German company compared to its Dutch rival to the north, I doubt this state of affairs will last more than another 3 quarters. Whereas Infineon's M&A plan seems to be one of consolidation on proven company strengths, NXP is looking to bolster its positioning for future market opportunities - in particular, the IoT. The company's acquisition of Freescale brings significant added capability in MCU and DSP while also giving NXP a large local presence in North America. When the merger is completed (perhaps by Q4), the combined enterprise will be the largest European-based chip company by far.

NXP is also stealing a page from its rival STMicro in seeking out partnerships - on this occasion, in China and the USA to bolster its IoT positioning and expertise:

Fires that shook me once, but now to silent ashes fall'n away.
Cold upon the dead volcano sleeps the gleam of dying day. - Alfred, Lord Tennyson

The 'sick man' of this trio, is, sadly, STMicro. In contrast to the vigor and aggressiveness of its Northern European competitors, the French-Italian conglomerate continues on a path of decay that began a full four years ago. What is most remarkable about this gradually unfolding train wreck is the completely lackadaisical posture of the company. There are a steady stream of announcements concerning new partnerships and alliances along with repeated assurances delivered in soothing tones that the firm's fortunes are proceeding 'according to plan and expectations.' In the meantime, STMicro's technology and revenues continue to putrify.

Like IBM, STMicro is a company with a legacy of technological excellence and managerial vision that has been commandeered by a gang of self-serving hacks with about as much understanding of High Technology as an Amish congregation. Thus, my prescription for fixing STMicro is the same as that for IBM - fire the bums who are running the company, immediately. Every day they linger on as executive managers is a day they do further damage and waste the talent and efforts of the rank and file.

The Vanara

"nisi impunitatis cupido retinuisset, magnis semper conatibus adversa."
The desire for safety stands against every great and noble enterprise. - Tacitus

The legendary shape shifters of Vedic scripture, Hanuman being a leading example among them, are often depicted as adventurous spirits righting wrongs and bringing balance to a world in chaos. What defined these mythic beings foremost was their sense of purpose - helping the Gods in their efforts to guide and assist mankind, smiting demons and doing good deeds.

Yet though the companies do share some of the characteristics of their namesake mythical beings, a sense of direction and purpose seems to be largely missing, as evidenced by the latest financial results and historical revenue trends.

The Programmable Logic sector has been a bastion of limited competition and 60 point gross margins for three decades. The up front investment in chip and toolchain development along with the costs of a large applications support organization have created barriers to entry for this segment that have been sufficient to discourage all of the other major chip houses from entering the fray, despite the profitability of the sector.

As a consequence, competition between participants has been, for the most part, very unadventurous and unimaginative. Customers are so familiar by now with each company's offerings that they can switch between them with relative ease. For the past 15 years and more, what Xilinx and Altera have been relying on is their ability to move feature-tweaked architectures to the next process node and hope that the other guy slips up either in their hardware or tools development schedules.

Complicating matters is that this lack of competitive challenge from other silicon houses has inculcated executive management teams with a smug sense of technological superiority - as if they were somehow favored by divine or otherwise supernatural forces. Hence their religious devotion to bit level programmability as the only 'proper' way to design microelectronic solutions. This myopia has also stunted any serious commitment to innovation and creativity such that the market and application characteristics of the business have hardly changed since the industry's inception. Consequently, system designers almost always restrict their use of FPGAs and CPLDs to rather peripheral functions such as IP prototyping, system level 'bandaids' for bridging, interfacing and off-chip bug fixing, or low volume pilot production runs of 1,000 to 20,000 units. It is almost unheard of to see a Xilinx or Altera product acting as the centerpiece of a medium to high volume system design.

This combination of complacency, inertia, lack of rigorous competitive stimulus and rigid conformance to doctrine is evident in the 8 year revenue history of both Xilinx and Altera. Earnings have been on a gradually declining slope for both companies since the second half of 2010, with any anomalous spurts in income for one or the other lasting only for two or three quarters before the revenue curves reverted to normal. 

There are a few news items of passing interest for both of the programmable logic leaders. Altera still has rumors swirling around it as a potential acquisition target by Intel. Some of the arguments for such a merger include diversification for Intel, technology growth from word level to bit level programmability, and integration of programmable logic into the Intel datacenter business, which would serve as reinforcement for what is already an Intel success story. Arguments against include the likely exorbitant price tag for the acquisition, the potential distraction and loss of focus for Intel from absorbing a business which is focused on markets where Intel has little or no presence, the current unsuitability of Altera's programmable logic offerings for emerging markets such as the IoT and the question of why Intel doesn't simply craft the capability internally, as the company has a history of developing and marketing its own PLD products. For my part, I am not really in one camp or the other and have not made up my mind. I'm open to any arguments that readers are interested in presenting on this issue.

As regards Xilinx, the most interesting bit of news lately has been its announcement of the 16nm Ultrascale FPGA family.

To put it simply, it currently appears that the Ultrascale family will incorporate very large tracts of configurable logic cells & I/O, a multi-core ARM CPU and  a huge assortment of hardware accelerators, memory, standards-based embedded IP, peripherals and a hierarchical bus so that a device could be used to prototype nearly any communications, computing, consumer, industrial, medical or automotive application that anyone could think of.

Xilinx has always had superlative design talent. It is the company's design engineers and FAE team that have made Xilinx successful despite the shenanigans of multiple waves of executive management. I have no doubt that the Ultrascale devices will be very impressive technically once they are released to production. 

However, at this moment the Ultrascale family appears to be a very expensive science project. It would be very surprising if such devices, with all their extra capability, didn't prove to be simply too expensive and over-featured for the great majority of system design projects, even if used just for prototyping purposes. If the goal of this product family is to help Xilinx learn exactly what sorts of IP and SoC capabilities it needs to develop in the future for products that intelligently mix programmable and fixed function circuitry for specific markets and applications, it seems like an awfully expensive way of doing this research.

In valor you are their equals; in necessity, the last and strongest weapon, their superiors. - Livy

The only programmable logic company that seems to have a clearly defined sense of purpose and mission is still the smallest one - Lattice. As discussed in last quarter's "State of the Union", Lattice is painstakingly carving out a position for itself in the IoT by offering I/O and low power circuitry (both programmable and fixed function) that is tailored for the demands of IoT system applications. The company evidently has a clear understanding of the technology it must develop or acquire in order to present a compelling value proposition to the budding IoT market.

The revenue numbers for all three companies illustrate the evident differences between Lattice and the two current market leaders. Notice how Lattice revenues have stayed rather steady over the last 5-6 quarters. Though this is not as desirable as showing 6 quarters of strong growth, it does compare quite favorably to the revenue track of both Altera and Xilinx, who are in particular showing an accelerating revenue decline over the last three quarters - a worrisome sign for both companies and a highly foreboding one for the High Tech industry as a whole (though that is a topic for next week's post.)

The Stone Masons

Source: (The Fountain of Neptune, Piazza Maggiore, Bologna, Italy)

The stone unhewn and cold
Becomes a living mould,
The more the marble wastes
The more the statue grows. - Michelangelo

Finally, we come to the 'cream of the crop' of the semiconductor industry - the sculptors who create wonders of functional beauty from plain, ordinary stone. Over the last two decades, the combination of talent and daring at these companies transferred significant value from system level hardware and software down to the chip level. Their IP portfolios and technology menus are the deepest and richest in the industry.

Yet the winds of change blow forcefully, suddenly and unpredictably across the High Tech seas, testing even the ablest of navigators. How well are the Ships of the Line of the semiconductor industry faring with today's winds and waves?

For convenience's sake, we'll go thru these companies alphabetically:

Broadcom - Very few High Tech firms could claim to work as hard as the rank & file do at Broadcom, and none could convincingly argue that they work harder. It's a very seasoned and extraordinarily talented group with an enormous latent capacity for invention and creativity. Sadly, though, it is mostly untapped by a command & control obsessed executive management.

As discussed in previous reports, Intel has quietly painted a bullseye on Broadcom and has made moves (including acquisitions) to challenge it in the STB sector - a Broadcom citadel. Rather than selling off this business unit as it has often elected to do with other BUs in the past, though, Broadcom is trying to hit back:

Nevertheless, despite one of the absolute best IP portfolios in High Tech, the crop Broadcom's executives have sewn with their oppressively dictatorial management practices is beginning to yield its predictably bitter harvest. Revenues are, after a period of relative stagnation going back to the second half of 2010, clearly on a downward slide. Past history and Broadcom's press announcements going back into 2014 suggest that the current executive management does not know how to reverse this and build growth momentum, as they do not know how to productively tap into the value-creating energy of their staff.

"De l'audace, encore de l'audace, toujours de l'audace."
Audacity, more audacity, always audacity. - Georges Jacques Danton, early leader of the French Revolution


Mediatek - After growth in 2013 and the first half of 2014 (strongly driven by the MStar acquisition), this company is showing the same extended earnings decline. However, there is hidden strength in the company, in particular for the low cost baseband cellphone market in China, where Mediatek is doing its best to resist pressure from Marvell and mighty Qualcomm. Furthermore, executive management appears to be decisive and aggressive, especially as regards new and developing market opportunities:

The company is also showing great fearlessness by stepping directly into Qualcomm's sights in the high end smartphone applications processor market:

Boldness in the face of adversity is a true demonstration of courage and resolve. Mediatek may very well switch places with Broadcom over the next 18-24 months and will bear watching.

Nvidia - This company seems to be the one constant in an ever-changing High Tech universe. It's cash cow GPU business cushions the company from major shocks. Those cushions appear to be a bit too comfy, though, as the company has yet to achieve any lasting success in new markets. The Tegra failed against Qualcomm in the mobile applications processor space despite a very promising start, and now Nvidia is withdrawing completely from the baseband market:

The company has poured enormous resources (both financial and engineering) into applied R&D for Robotics, AI, Machine Vision and Voice Recognition/Activation. Yet the effect on the bottom line has been negligible at best. I know some of the engineering and research talent at the company; the fact that their work has not resulted in oceans of income for Nvidia is very revealing.

The company is notorious for a management style no less callous and oppressive than Broadcom's. The results are, of course, the same: an inability on the part of the executive team to capitalize on the enormous wealth of talent at their disposal in the rank & file.

One day, investors will tire of this state of affairs and begin to move against management in the way that Google's investors have begun to do (as was discussed last week.) When that day will come is impossible to say, though it will most certainly arrive.

Qualcomm - This company continues to impress as a leader of leaders in the chip industry. While other top SoC vendors have been shrinking or stagnating, Qualcomm - in the face of continuing competition against both its baseband and applications processor chips in a saturating smartphone market stifled by chronic consumer spending weakness- still finds ways to grow.

Its success has attracted the attention of some truly formidable enemies, as the $975M fine by a China regulatory agency for 'anti-competitive practices' really amounted to nothing more than a coercive extortion by Beijing of price breaks for local phone manufacturers:

Yet the company is able to take its short term troubles in stride and still execute on its long term vision for mobile computing - a vision including a level of clarity and daring that rivals even Apple's:,28670.html

Qualcomm's conception of a personal AI residing on individual smartphones/portable processors is complemented by its farsightedness in the semiconductor technology that will be needed to support that goal - a method for implementing true monolithic 3D-IC that is ingenious for its very simplicity (at least in concept, though there are of course considerable engineering challenges to overcome for mass production:)

The company is uniquely blessed, in that it has both second-to-none talent and an extraordinary executive management team. I envy and admire the folks who work at Qualcomm and expect to see their success continue.


Now that we have covered all the companies in the portfolio, we are left with a final question: where is the High Tech industry going from here?

We'll discuss this next week as we look at broader economic factors that will affect the Technology industry in 2015 and beyond. Thus, once again we will be sailing our vessel into port and travelling overland to the province of Boeotia in Ancient Greece, climbing the slopes of Mount Parnassus and leaving an offering at the Temple of Apollo, as we seek an audience with the Pythia to receive the guidance of prophecy from high Olympus...........


Open Disclosure: I own no stock in any reviewed company and am not being paid by them or any of their agents to write any opinions, good or bad.
Dear readers,
Don't forget the Amazon and FlipKart banners! Many thanks to those who already have clicked thru them to do their shopping, and I hope more will join in to help keep this blog going. :-)
p.s. If any of you would like those banners to hilite particular products or categories, please let me know.

No comments:

Post a Comment

Feel free to comment or critique!