Someone asked me if there is something wrong with the semiconductor industry. Like companies like SiliconBlue and PicoChip that look good and have excellent technology, why would they sell them at a lower price than the few venture capital investments they have received in years? I don't think that there is any fundamental mistake in the semiconductor industry. The industry is currently only entering a market and technology, and the reorganization stage that investors are looking forward to, as Moore's Law is going downhill and is coming to an end. Trying to transform.

In the past, there were many large companies that used Moore's Law to achieve leading product performance and reduce costs to outperform their rivals; this worked in the past, starting from the period when all the wafer manufacturers produced similar functional block components. Therefore, for the wafer manufacturers at that time, the added value does not lie in the design, but in the manufacturing expertise.

Indeed, a few years ago, Moore's law's process miniaturization results always seem to be economically successful - how quickly integrated component manufacturers (IDMs) can move to next-generation process nodes, and a single component will be able to How quickly the design case is obtained at a lower price also means how quickly competitors can design competitive products. At the same time, both the global electronics industry and the semiconductor market are growing at multiples.

But now, we are very close to the terminus of Moore's Law, and the cost of non-recurring engineering (NRE) for product R&D and production has far exceeded the economic benefits that can be achieved by investing in advanced process nodes. The few exceptions are those very mass-produced components such as memory, processors, FPGAs, and some very demanding system chips.

Another major change is that for IDM vendors and fabless chip makers, the added value now comes from the design more than manufacturing. At present, most of the wafer manufacturers have more or less reached the same CMOS process node, so to achieve a difference in sales results, is a better design, or at least get a large customer design case, to ensure that the start of the chip Shipments.

From another point of view, Moore's Law can provide a reduction in functional advantages and unit cost advantages, but the required NRE cost is multiplied. For most companies, it is time to step down Moore's Law. Escalator." It is not surprising that investing $1,000 to $50 million to sell a company for $250 million, because now our cost and transaction value have been reversed. Some venture capitalists have spent 100 to 500 million U.S. dollars but have only recovered 50 million U.S. dollars. It is no wonder that these venture capitalists have turned around in recent years instead of continuing to focus on similar investment opportunities.

The good news is that the solution to this situation is clear and has been accepted by semiconductor manufacturers worldwide; some people call it "more-than-Moore." In 2011, The International Technology Roadmap for Semiconductors (ITRS) published an excellent white paper stating that some applications, including medical care, transportation and energy, would benefit from "surpassing Moore's Law." Instead of process shrinking.

Although ITRS said that other applications in security, communications, automotive infotainment, etc. can also benefit from the billions of transistors on a single chip, it is clear that only one or two very few companies can Lead a large number of chips that are sufficient to balance the required cost. Therefore, for most start-up companies, the key is to prepare to get rid of Moore's Law and free to leave.

Why not give yourself a cost advantage of not using advanced process technology anymore, and don't always try or push some commonly-used communication standards on digital chips, etc.; all in all, traditional innovation comes from smart design, and System-level functionality has been completely improved. This is the innovation that MEMS inertial sensors and navigation elements made by companies such as InvenSense (recently IPO), and Nintendo, the console manufacturer, brought surprise to the world with Wii somatosensory control.

It is also worth noting that there are many MEMS manufacturers can call for a good purchase price. Murata Manufacturing, for example, recently announced that it has agreed to spend 20 billion yen (approximately 190 million euros or 260 million US dollars) to acquire Finnish company VTI Technologies; and Maxim Integrated Products to bring SensorDynamics under its umbrella for 130 million US dollars.

We really can't see the same scale of venture capital investment that happened in the first half of the past decade, but the good ideas will still be supported. This is because, in any case, in its widest form of expression, silicon, semiconductors, semiconductors And electronic technology is still the best way to create wealth and improve the well-being of humanity throughout the world.