Internet Giants and the Need for Faster AI: Part 2
Yesterday, I began this discussion of the need for speed when it comes to artificial intelligence (AI). If you missed it, you can click here to catch up.
Now, Alibaba got into the business of making its custom AI chips by acquiring an ASIC (application specific integrated circuit) maker called C-Sky, which it renamed Pingtogue. Pintogue now drives all semiconductor development for Alibaba, a way of ensuring that Pintogue will develop the chips Alibaba needs. That’s integration.
However, buying capabilities is not the only path to technology integration. After all, Pintogue itself does not incorporate all the competencies required to build the array of AI chips Alibaba will need. Among those it is missing is: competency in processors whose logic paths can be reset in the field, based on experience, often known as field programmable gate arrays (FPGAs).
The Need for Adaptable Chips
At its core, artificial intelligence is a learning machine, continuously adapting and course-correcting as it encounters new data. This requires that the chips themselves also be adaptable. Programmability is particularly important in AI applications because the algorithms and neural networks are constantly evolving, getting better, faster, more efficient, and more accurate day-by-day.
Pintogue’s specialty is building application specific integrated circuits (ASICs) chips for custom applications. This includes leading-edge AI performance, but once an ASIC is created its logic functions — however customized — are locked in. If an ASIC needs updating, tweaking, or bug fixing it means a new chip, new design, new fabrication line. It’s not only back to the drawing board, it’s back to the bank.
This cost and lack of flexibility are big reasons for the tremendous appeal of FPGAs (produced by the likes of Xilinx, Intel, and Achronix). FPGAs include logic gates that are not specified at the factory. These chips can be electronically re-programmed in the field so that, for instance, a piece of “and” logic becomes an “or” or a “nor” or a “nand.” Transistors that pass current can be re-programmed to block it.
FPGAs can accommodate virtually any logic design, from networking to artificial intelligence, and data signal processing. They can be quite expensive on a per-unit basis, but the non-recurring engineering costs involve only design efforts and not the costly fabrication (and long schedule times) of a fully custom ASIC device.
Heretofore, engineers had to choose: ASICs, cheap in volume, but fixed in function? Or FPGAs, individually more expensive but flexible.
Why not combine them instead?
The Benefit of Programmable ASICs and the IP Ecosystem Alliance
Once considered a pipedream, programmable ASICs are just about to hit the streets
Making the ASIC programmable is an embedded FPGA core integrated into the ASIC design. It combines the benefits of programmable logic with the low recurring cost of an ASIC. And it can improve performance parameters such as latency, throughput, and power.
By now I’m sure you’re thinking “OK, George. That’s impressive. But are you ever going to get back to Alibaba, or even China? We are feeling a little like Marco Polo: The Lost Years right at the moment.”
Actually, I’ve never left China. Remember how we said Pintogue didn’t have FPGA capabilities? Now it does because it has just created its “IP Ecosystem Alliance.” And the company that was invited into that alliance to create field programmable ASICs is a tiny American company we recommended in the most recent edition of George Gilder’s Moonshots. Click here to access the company name
Such alliances are another way of going at integration: build ongoing partnerships with firms with capabilities you lack so you can design products and build IP together. It’s not exactly marriage, but it beats running out to the hardware store every time you need something.
The moral of the story reaches far beyond this one American company. Actually, we have two morals.
Moral #1: The first is that competitions for leadership slots in the internet economy will often turn on the hardest of hardcore technologies, even if the competitors in question look like “soft” consumer-oriented businesses. When evaluating even consumer companies, get beyond the chrome and look under the hood.
Moral #2: Even more important, it doesn’t matter how much the politicians want to turn the world economy into something they imagine they understand, like wars. (Have you noticed how they are always declaring “war” on something.) Today even more than in Adam Smith’s day, there are no final victors, and beggar thy neighbor policies beggar everybody.
The advances of the mind, whether they happen in the US, China, Israel, South Africa, or Sweden — are advances for us all. And the creators will strive to find each other no matter how much governments want to keep them apart.
Editor, Gilder’s Daily Prophecy