World Domination Through Vertical Integration

January 7, 2011
Russell Fish

Vertical Integration is the control of two or more levels in the supply chain of a product.  For instance, a clothing manufacturer might acquire or start a textile maker.  The clothing maker could exclusively source cloth that would be unavailable to competitors.  Similarly an auto maker might acquire a steel maker that produced a very strong alloy.  The auto maker could integrate a step further and acquire mines.

Vertical Integration makes sense as a strategy only when an upstream supplier possesses some unique product or capability.  In such a case, the uniqueness of the upstream supplier manifests itself as uniqueness in the downstream product.  The value of the upstream uniqueness is amplified by the relative cost of the upstream raw material to the downstream end product.

For example, some Chinese companies are attempting to create a vertically integrated supply chain that begins with rare earth ores such as Bastnäsite of which China dominates commercial world reserves.    Bastnäsite can be refined to neodymium which is then alloyed with iron and boron.  Magnets produced from this alloy are able to produce magnetic fields more than twice the strength of similar sized magnets made from other materials.  Neodymium magnets are unique.

The Chinese companies plan to design these unique magnets into DC motors.  These motors are far more efficient and lighter weight than non-neodymium motors.

They further intend to build the motors into electric vehicles.  Such vehicles will run 20% farther on the same electrical energy as non-neodymium vehicles.  Any other companies will compete at a distinct disadvantage.

If this vertical integration chain is achieved, one dollar of Bastnäsite ore will control $100 of electric vehicle revenue.

Apple Makes Sauce of the Personal Computer Industry

Vertical integration has been mostly absent through the history of the personal computer.

The industry was built on standardized or commoditized components.  The IBM PC established the hardware standard for CPU, IO, busses, and operating system.  The CPU and operating system were standardized on INTEL and Microsoft.  Microsoft never had direct competition, and INTEL only had token competition from AMD.  The other components, e.g. disk drives, memory, video controllers, power supplies, etc. were multi-sourced commodities.  No PC company attempted a vertical integration strategy... until Apple.

In 2008 Apple acquired Palo Alto Semiconductor, a microprocessor design house headed by computer architect Daniel W. Dobberpuhl.  As part of the acquisition they acquired a portfolio of patents including:

7,652,494, Operating an integrated circuit at a minimum supply voltage

This patent adjusted the operating voltage of a microprocessor to match the operating frequency.  The result was potentially dramatic power savings over chips that do not have this feature.  Apple also acquired an ARM license that allowed them to re-implement the architecture in a unique way.

On January 27, 2010, Apple introduced the iPad tablet computer running a proprietary CPU called the A4.  The world was stunned at its slim profile and greater than 10 hour operating time between charges.  The largest computer company in the world rushed out the HP Windows 7 Slate.  With a 5 hour operating time the Slate was dead on arrival.

On June 7, Apple announced that the iPhone 4 would use the A4.  On September 1, they announced that Apple TV would be upgraded to run the A4 CPU.  It was now clear that Apple would include the A4 and its derivatives wherever it could gain advantage.

The CPU was less than 10% of the cost of goods of an iPhone, but it controlled the strategic advantage of the entire product.  Steve had executed a successful vertical integration including CPU, operating system, end computing product, North Carolina Apple Data Center, and Apple Store retailer.

Industry Response

When competitors first glimpsed the iPad, the initial reaction was, "Damn it's thin.  They must have scrimped on the battery."  When the first reviewers reported battery lifetime in excess of 10 hours, panic set in.  It was clear that Steve had the lowest power CPU in the industry and a two year lead over everybody else.  The industry went on a CPU acquisition spree.

Steve's buddy Larry Ellison already had a vertical integration plan he called "The Big Red Stack".  The same day as the iPad announcement Oracle acquired Sun and its SPARC CPU division for $7 billion.  Larry was soon bragging how he would crush HP with hardware.

Three months later Google bought Agnilux, a stealth CPU company purported to be developing lower power chips.  Agnilux consisted of former PA Semi designers.

Three months after that Microsoft acquired an ARM "architectural license" that allowed them to design their own ARM chips.  After two decades as one half of the Wintel alliance, Microsoft was rumored to be preparing a Windows version to run on the ARM architecture.

The biggest loser was of course INTEL.  In one morning, Apple destroyed the Netbook business on which INTEL had bet several 100 million dollars in developing the ATOM.  ATOM consumed more than ten times the power of the A4.  The future of computers was going to be low power tablets and low power cloud servers for which INTEL had only power hungry legacy solutions.

Within 7 months INTEL bought Infineon's ARM making division for $1.4 billion which also included an "architectural license".

TOMI™ the Milliwatt Microprocessor

As of the end of 2010, Steve Jobs owned the future of the computer business.  He had at least a two year lead over his closest competitors, and he had a strong CPU power reduction patent portfolio.  He was also most assuredly not standing still.

Whatever competitors develop today must compete not with today's A4, but with whatever new and better CPU comes out of Cupertino.  This is reminiscent of the grinding existence of AMD chasing INTEL for 20 years.  Being number 2 really sucks.

To catch Apple, a competitor must jump ahead.

However fundamental limitations on legacy computer architectures were encapsulated by Dr. David Patterson in "Patterson's Three Walls":

Power Wall + Memory Wall + ILP Wall = Brick Wall

Apple had aggressively attacked the Power Wall with its proprietary IP to gain as much as a 2X power advantage over its closest ARM rivals.  This didn't leave much room for a competitor to improve on a legacy computer architecture.

There may be a new option.

An obscure design group from Dallas claims they have broken Patterson's Walls by refactoring the division between CPU and main memory.  They call it TOMI™ Technology and claim to put a quad-core CPU in the middle of a commodity DRAM chip.  The prototype designed on a turn-of-the-century 110nm 64M DRAM clocks at 500MHz while consuming 23mw per CPU.

Furthermore, according to Microprocessor Report, "Venrays prototype can deliver 4.6 times the bandwidth of Intels best server processor despite using 10-year-old DRAM technology."

The 64M parent DRAM sells for 65 cents.  The 4 CPUs add only 20% to the die size of the DRAM so the new technology is fairly cost effective.

Maybe the Chinese can figure out how to combine TOMI™ with their rare earth magnets.