Innovation and the Business Cycle

There needs to be a match between the innovation cycle and business cycle, that is, the cadence at which each develops. In one company I worked for, the business time scale was set at two years, which means you need to go from concept to product shipment in that time. But the innovation time scale for silicon photonics, which was my team’s focus, is also just about two years, so this was a tight fit.

Risk is the driving concept here. There’s often lots of interest from senior management about failing fast, being willing to take risks, etc. What they mean is that they are comfortable with taking calculated risks, doing experiments to learn, even if it takes money, resources, etc., but that those risks need to resolve on a certain timescale. You can't take a risk today and not find out the results in two years, or even one year. This is tough for silicon photonics, where risks last 6 months, 8 months, even 1 year - based on the time scale of design, tape out, manufacture, measure, and analyze.

We can introduce the concept of risk duration, which is how long a risks take to resolve. This is different from the probability of success, the cost of failure or reward for success, and the sunk cost of doing the experiment. Risk duration is intimately related to the cadence of innovation, or the innovation timescale, of a technology or platform. To best exploit a technology, its risk duration should be quite a bit smaller than the business cycle of the company or the industry it's in.

That's a key challenge in for silicon photonics. Most companies are willing to risk money today to drive innovation for the future, but there’s a limit to how for in the future. Part of the reason is market risk. Markets change fast, so even if a technical risk works out, the reward is contingent on the market still being there. And that is arguably harder to predict than whether the technical risk will work out.