Mad Dog 21/21: Newtonian Economics
Published: August 27, 2008
by Hesh Wiener
Isaac Newton's contributions to science and mathematics were monumental. They elevated English technology to a state of great prominence and helped usher in the Age of Enlightenment. More than three centuries later, every teenager studying science is taught classical mechanics, and Newton's principles provides food for thought in other disciplines, too. IBM's recent financial results, a sunny patch on a graying economic landscape, beg for analysis as well as praise, and Newton's laws might offer just the metaphorical basis required to examine Big Blue's blooming.
The top stories in the business press may be dire, but as we previously reported here at IT Jungle, IBM reported a very good second quarter: revenue climbed 13 percent to $27 billion and profit went up 22 percent to $2.8 billion. Because IBM keeps buying back its own shares, the figures on a per-share basis were even more impressive.
IBM is not only bucking the aggregate trend, it seems to be doing well smack in the middle of the storm. A big slice of IBM's revenue, approximately 29 percent, comes from customers in finance. Quite a few of these IBM customers are reporting terrible financial results, but they are apparently still paying plenty for hardware, software, and services. About 81 percent of IBM's revenue across the board comes from big business and government.
IBM is not dependent on small businesses to the same extent as Hewlett-Packard and Dell, the leaders in the PC and X64 server markets. Where IBM is trying to serve small business with proprietary midrange systems, competing with X64, Unix, and Linux platforms, it is getting less satisfactory results. But Big Blue's current frustration in the SMB market could be temporary. The company is amidst a transition in its midrange product lines and its strategy has quite a bit in common with the process that led to reinvigoration of its trade with big business.
Isaac Newton: Arguably the most brilliant among a group of scientists, mathematicians, inventors, architects, and engineers who led England's rise as an intellectual power in the 17th century.
What seems to bring IBM the best results is similar to the effect that brought excellent results to Isaac Newton: combining hardware and intangibles in a way that multiplies rather than merely adds the benefits of each component.
In Newton's case, the intangible work was a theory of universal gravitation that was explained in the mathematician's masterpiece, Philosophiae Naturalis Principia Mathematica, commonly called just Principia. Newton's work described with considerable accuracy the motion of the planets in our solar system and even led to a prediction, when Uranus was discovered to be a planet (about 50 years after Newton's death) that there was a yet-another planet beyond it, shaping its orbit.
One of the problems Newton and his contemporaries had when it came to examining the heavens was the limited quality of large lenses used in telescopes. As Newton himself helped discover, glass lenses suffer from a distortion called chromatic aberration, which means that light of different colors is refracted at different angles. Newton found a way around this problem. He invented the reflecting telescope, a device that significantly improved the quality of images available to astronomers and is still the basis of nearly all large optical telescopes.
In IBM's case, the challenge of selling ever more expensive central systems to big customers while hardware price/performance keeps improving is a serious matter. In markets outside the mainframe realm, there seems to be essentially unlimited demand for computing power and the size of the market grows faster than the cost per unit of processing power falls. This is something economists call positive price elasticity and it is the basis of success of industrial production in many markets. But IBM's marketing model for mainframes involves a mix of hardware, software, maintenance, and services because the hardware market in isolation is apparently not sufficiently elastic. If mainframe price/performance fell as quickly as it does in the case of smaller servers, sales each year would diminish very rapidly, or so IBM appears to think.
Demand for mainframe power seems to keep increasing, but the demand is due to causes other than improved price/performance. In the most recent quarter, IBM's mainframe sales rose by an impressive 32 percent. MIPS shipments went up 34 percent. That means IBM probably did not get the jump in sales by reducing its average cost per MIPS.
It must be pointed out that IBM charges more per MIPS on smaller mainframe systems, so it could have achieved the reported results by cutting prices on machines of all sizes and at the same time bringing in more business from buyers of relatively low-powered mainframes, but we doubt that is the case. IBM seems to be doing well with its new z10 machines, which for now only come in large sizes. Customers who want mainframes less powerful than the entry level z10 machines have to buy z9 boxes, which are well along in their product cycle, when customer resistance begins to arise. IBM is apparently getting satisfactory sales results from the z9, but it is doubtful whether total sales of low end versions is sufficient to produce the dollar and MIPS shipment results IBM has reported.
The upshot is that IBM's market model for big customers buying big iron is based on the belief that the only way to bring in continual growth is to add to the scope of goods and services sold at each account. Moreover, as the population of small mainframes declines, which it has done for years, the base of servers in midsize companies keeps rising, so the issue is not whether companies who don't need very large computers are buying more servers, because they are, but rather that they seem inclined to migrate off mainframes for a number of reasons that include technical considerations, economic factors, and management issues. Small mainframes may have impressive credentials, but they are hard to sell.
In the proprietary midrange market, where IBM offers its Power Systems servers and i 6.1 operating system, the most recent quarter's results were looked just awful on the face of it because IBM has moved these Power-based server sales into a converged System p category for reporting purposes. And therefore, sales of what IBM calls its legacy System i servers--which means any i box not based on Power6 iron--fell to just over half the revenue IBM achieved last year, and last year IBM reported a 15 percent drop from the prior year's second quarter volume. IBM did not report a combined i platform revenue figure, but elsewhere in this newsletter, we present our best guess about how real i and p server sales came in at the end of Q2.
IBM is widely expected to get improved results in the i product line, but that improved outcome is only now beginning to become visible. IBM has eliminated remaining hardware and pricing differences between i and p series servers and is revamping its marketing around the resultant Power Systems line. There are at least two major forces in this move that IBM could use to stimulate sales. One is the improved value the new machines will provide to the extent IBM wishes to move i boxes closer in price/performance to similar p machines. The other is the improved versatility the new machines will provide, allowing customers to take even greater advantage of Linux, Unix, and Windows applications than they can today. Here, too, the magnitude of the force will depend on IBM's marketing decisions. IBM is aware that frustrated i users will not simply switch to other IBM platforms; they will change vendors. As is the case in the small mainframe segment, once users defect they rarely return.
IBM is optimistic about the possibilities of a renovated i line because it has done so well with improved Power Unix servers. The converged System p business was up 29 percent year-to-year during the second quarter, and a lot of that that rise is due to Power6 successors to the legacy System i machines that IBM is phasing out of production. (To finish my overview of IBM's server business, I should mention that the company's trade in X64 boxes was off by 5 percent in the quarter. IBM is not limber enough to do well in commodity markets, just as successful vendors of commodity systems have a hard time selling richer and more complex packages of hardware, software, and services.)
Newton's Principia: In this three-volume book Newton published his version of calculus, his thesis of classical mechanics, and the analysis that explained the shape of planetary orbits.
Observers who doubt IBM can ultimately rescue the proprietary mainframe and midrange markets from the growing (in every sense) scope of standardized servers believe that the recent past points to the near future. Those who believe IBM will gain ground in the current climate of risk-averse purchasing feel that Big Blue is doing the work required to change direction.
Newton's first laws of motion says, more or less, that every object in a state of uniform motion (or which is not moving at all) will remain in motion unless some force acts on it (or remain still). In other worse, inertia and momentum don't change by themselves; they change only in response to a force. Viewed from the outside, based on gross aggregate figures, it is reasonable to guess that mainframe and Power Systems sales will rise in the future and that by the end of the year the legacy System i market will be quite small as customers are increasingly tempted by the next generation of IBM midrange technology. It's also very possibly a safe bet that IBM's X64 business will be soft or maybe worse.
IBM's software and services business, which brings Big Blue half its intake, is growing about half as fast as mainframe hardware sales and quite a lot faster than IBM's total hardware sales (which rose only 2 percent in the quarter and is still running a shade under last year's sales rate for the first half). IBM's hardware sales have not only been hurt by headwinds in the small and medium business markets, but also by unexpectedly slow sales of the chips IBM makes for use in other companies' game consoles.
Despite the mixed picture, IBM has predicted its earnings per share for the full year will be about 25 cents higher than it at first thought, hitting $8.75 rather than the original estimate of $8.50. That kind of result would be 22 percent better than the EPS IBM reported for all of last year.
In order to do this, IBM might have to get some of its activities to grow faster or, as they say in mechanics, to accelerate. Newton, in his second law of motion, said that acceleration is connected to the mass of an object and the force that is applied to it. Specifically, force is mass times acceleration or, as we are talking about acceleration here, acceleration is force divided by mass. Both acceleration and force are vectors, meaning they have magnitude and direction. So, if Newton were alive and putting his sterling into IBM shares, he would be betting that IBM not only is going to give its various business activities a hefty shove, but also that it would shove things in the right direction. The shove has to be hefty in a Newtonian world because IBM is so massive.
IBM can't use the same kind of shoving in the big business segment that it used to use when it made PCs and that it sometimes uses today as it peddles midrange and small systems directly or though dealers. The kind of shoving IBM does in the small ticket market is based to a considerable extent on advertising. Advertising generally works better if the buyer knows what the seller is offering. Customers know what a desktop or laptop machine is and they have a pretty accurate conception of a small server. Mainframes are another thing entirely. IBM has made the mainframe so complicated with specialty engines, Byzantine software pricing plans, elaborate lease and upgrade schemes, and deals that move functionality from z/OS to Linux at the drop of a budget.
Some of what appears to be a perpetual migration to more cost-effective mainframe technologies turns out to be an Escher staircase. Customers can see this when they check their bills and discover just how it is that IBM seems to be growing so nicely even when hard times are forcing so many customers to press for reduced costs and, in fact, get offers that seem likely to deliver savings. Lots of items may become cheaper but the total package always seems to grow.
As big business increasingly turns to big packages of hardware, software, and services and consequently depends more on IBM (or whoever it turns to for the deal) and less on its own personnel, the customer loses control. Sure, a company can decide to sign or not sign, it can try to manage what it buys, and it can bring in competitors whenever contracts lapse, but the big picture for big business is that customers can make strategic choices but no longer can control tactics.
Newton's Telescope: When his work in optics showed that telescope lenses produced chromatic aberrations, Newton developed the reflecting telescope that was used to support his analysis of motion in the solar system.
This can be great for some customers and it can be not so great for others. It comes down to cases. Whatever the results at the outset of a transition to services, in the long run it is clear that a customer quickly loses the ability to switch a job from an expensive platform to a cheaper one, or to get savings passed through when a services provider makes such a switch. Platform selection is a privilege reserved for user organizations that have found a way to preserve control over both tactics and strategy.
For big business, in many ways there are fewer choices than you find in the midrange or low end of the marketplace. Big iron is mainly mainframes, even though Unix boxes, including the ones offered by IBM, are just as reliable and often more powerful. They are also cheaper. But just as IBM has to exert a lot of power to change its direction, so, too, does a user. It's expensive, frustrating, and sometimes completely impractical for a customer to move work from a mainframe to a Linux box, and arguably even more difficult when the Linux box is not built into a legacy IBM frame.
To be fair, it is also worth observing that big businesses that use X64 technology, even Google with all its smart techies, might find it hard to move from Linux or BSD Unix to AIX let alone z/OS or what IBM now calls the i operating system (and many users still think of as OS/400). As Newton's work points out, the rules of mechanics apply to all objects. The same kind of universality arises under the rules of economics. A force such as friction affects the motion of objects less when the objects are really massive and the friction is relatively small (the way air resistance is small compared to the force of a flying cannonball). In business situations, a company with tons of mainframes, one with tons of X64 boxes, one with tons of Solaris machines, or one with tons of HP Itanium hardware is going to have a lot of strategic momentum. It's not the MIPS; it's the motion.
Still the processes used by IBM's managers to alter the company's motion are not without their limitations, including the way they are shifted in magnitude and direction as customers and competitors adjust to whatever IBM does.
Newton put it more or less this way: For every action there is an equal and opposite reaction.
The way Newton's Third Law works is easy to see when you step out of a small boat. You move toward the dock, or try to. The boat moves away from the dock. If you don't figure it right, you might end up in the drink.
It's harder to spot the same process when you jump into the air and are pulled back to Earth by gravity, but in fact you are pulling on the Earth just as hard as the Earth is pulling on you. It's just that you are dinky compared to the Earth, so it looks like the Earth hasn't moved at all.
As I said, IBM believes that its sales of goods and services to big business combined with its stock buybacks is going to boost its EPS to $8.75 this year. The company must be confident. It if revises its forecast downward, its stock would get whacked and some big heads could roll.
IBM's models predict great results based on what it plans to do, how much it expects to spend doing it, and what it can get from customers even in these tough times. IBM has also figured just how often it will have to adjust its deals to match competition, which is not an issue when it comes to mainframe hardware but is a factor in total packages. Some services companies might beat IBM in some deals, but if they have to buy IBM iron and license a lot of IBM software, Big Blue will make out anyway.
Where IBM can get badly hurt is in cases where a complete business function is moved to a system, software, and services offering that is entirely non-IBM. If HP-EDS can beat IBM with a package that includes migration, IBM would not only lose a deal, it would face a very severe challenge winning back what it lost. The cost and hassle of reverse migration might take IBM out of the running. The only way IBM could regain the deal is to offer to do just what its rival has done but do it cheaper and better.
Another possible problem can occur even when customers want the IBM package if they cannot afford it. IBM can either cut its prices or learn, as its customers may, how to manage with a Small Blue bundle rather than the Big Blue one in the pre-crash budget.
Still, sometimes things fall no matter what you do and the best you can get is an understanding of just what fell and why.
This is how Newton is said to have been led to his insights into the way gravity works: by the fall of an apple. Today, IBM might find that will have to adjust its expectations, as has happened quite recently when Apple was obliged to warn investors about the gravity of business conditions.
IBM Drives Home a Strong Second Quarter Across the Board
Intel Has a Great Q2, and AMD Has a Poor One and Taps a New CEO
Global Sales Save HP's Financial Cookies in the Second Quarter
Post this story to del.icio.us
Post this story to Digg
Post this story to Slashdot