No matter how microprocessor giant Intel (INTC +4.36%) tries to spin it, Moore's Law is dead.
For those unfamiliar with Moore's Law, it essentially says that every two years or so, the company would develop new chip manufacturing technologies that allowed its product development teams to cram in far more transistors (and therefore features) into new products while keeping overall product costs flat compared to prior generations (implying a reduction in the cost per transistor).
Image source: Intel.
Based on this definition of Moore's Law, Intel is no longer following it, even as it tries to assure investors otherwise.
Indeed, not only have transitions to newer manufacturing technologies lengthened for Intel (already violating the "law"), but even when Intel does transition to those newer technologies, manufacturing yields are so poor that product costs go up generation over generation at the beginning of a new manufacturing technology ramp-up.
The "death" of Moore's Law doesn't spell gloom and doom for the personal computer industry, however. There are certainly ways to build more feature-rich and powerful computer chips without relying on transistors getting smaller and cheaper, but it requires a fundamentally different corporate mind-set than the one that Intel seems to have had in the past.
According to recent commentary from Intel executive Murthy Renduchintala, it looks like the company is finally learning to accept the death of Moore's Law.
Focus on better products, not smaller transistors
"We're going to be focused more on the generation by the amount of performance increment it will give us," Renduchintala told PC World. "I don't think generations will be tagged to [manufacturing] node transitions."
Indeed, there are many ways for a chip company to deliver improved performance and features without needing to rely on transistors getting smaller.
For example, Intel's current seventh-generation Core processors are built on a performance-enhanced version of the company's original 14-nanometer technology, called 14-nanometer+. This technology doesn't provide an area reduction compared to the original 14-nanometer technology, but what that enhancement allowed Intel to do is deliver better performance and power efficiency than the sixth-generation Core processors built on the original 14-nanometer technology.
Going forward, it seems that Intel will continue to focus on trying to improve the underlying performance and power-efficiency characteristics of its pre-existing manufacturing technologies. Such improvements should allow the company to build increasingly better products without having to worry about the challenges associated with making transistors smaller (though it does need to worry about making those transistors better).
Furthermore, it's not all about manufacturing technology, either. Even without transitions to newer manufacturing technologies, Intel's chip design teams could make improvements to the underlying chip designs and architectures themselves.
Intel hasn't fully exploited that potential with its seventh-generation Core processors (the changes were mainly in the manufacturing technology an in reworking the circuit designs to use that technology), and it doesn't look like it will be doing much of that with its upcoming eighth-generation Core processors either (reliable leaks suggest that Intel will rely mainly on boosting processor core counts rather than making changes to the cores themselves).
However, in future generations -- now that the company is now explicitly planning around needing to use the same basic chip manufacturing technology (though with performance enhancements) -- Intel might have enough time to plan for more substantive chip design and architectural changes each year.
If Intel can manage to improve both its architectures and the performance characteristics of its manufacturing technologies at an annual clip, then the company should be in a good position to deliver significantly better products to its customers at a regular pace -- always a good thing.
