Competitors, partners, and customers in our interdependent semiconductor industry often have mutual interests that could benefit from cooperation. By offering customers efficiency, we'd all win. Sadly, that's not how things often work.
In the highly competitive market for chip design solutions, the system design ecosystem could do well with a bit of business ingenuity and a dash of the old retail adage, “The customer is always right.” This is especially true when a user company wants competitors or partners to work together to solve its efficiency challenges. Competitors, partners, and customers in our interdependent semiconductor industry often have mutual interests that could benefit from a cooperative, strategic relationship. Customer efficiency and giving the customer what it wants means we all win through economic growth.
Sadly, that’s not how things often work, even though the user community would appreciate consideration, cooperation, and not “winner takes all” at their expense.
Here’s an all-too-true example that’s often repeated today. The year was 1996, and the developer of a portfolio of scheduling products to reduce design costs, risk, and time-to-market was invited to participate in an event organized by a now-defunct computer manufacturer. A sales manager from a dominant vendor opposed its participation. His biggest fear, it was later discovered, came from the product’s ability to let users squeeze more cycles from existing licenses of the big vendor’s digital simulation software. Eventually, the computer company’s executives were able to soften his stance and we managed to participate in the event after all.
The sales manager’s initial position was proven wrong — added efficiency and license utilization meant an increase of the vendor’s revenue and that he better served his customer. While it’s easy to portray the sales manager as the “bad guy,” he was just doing his job. He wasn’t paid to think strategically.
History repeats itself 20 years later as a similar scenario plays out in another growing area. Hardware emulation, offered by several design automation companies, is gaining widespread adoption and increasing acceptance throughout the semiconductor industry. In many cases, it is underutilized despite engineering groups’ claims that their machines are fully booked and they need added capacity for future projects. A tool-agnostic piece of software called a job scheduler from another vendor could help streamline their jobs and increase overall efficiency. To the job scheduler, this particular tool appears as compute resources with some additional complications regarding what jobs can be executed concurrently.
While some companies may see this as a threat, it can be viewed as an opportunity as well. When efficiency goes up, customers capitalize on it by being able to complete more tasks in a shorter period of time, leading to faster time-to-market and/or the ability to finish more projects sooner. This leads to more investment, not less. By helping the customer achieve higher efficiency, the market “pie” grows.
Seeing a threat — give customers more efficiency and they won’t need to make any further investments — and not opportunity may seem logical, but it’s short-sighted. This might be true for a slow-moving and complacent market, but it is not the daily reality in the semiconductor market. Time-to-market is everything. Shaving a month off a product schedule is huge. Being able to complete three projects in the time required to complete one is even bigger. Efficiency makes a company more competitive.
Stories like these are repeated across the entire design system ecosystem as small and emerging companies appear threatening to bigger vendors. In reality, they could be cooperative partners growing tool usage and revenue, especially when two competitive vendors have similar interests to reach higher-value creation. The behavior does not serve the customer or engender a “the customer is always right” attitude, nor does it help the customer gain more efficient use of existing tools.
Here’s an example of an opportunity to grow the pie: A user had more than 100 hardware emulation licenses but was frustrated by an inability to use those licenses at more than 80% capacity due to the slowness of the job scheduler. A different job scheduler from another vendor was installed on the same machines. Using the same workload, the customer was right — the faster scheduler brought the license utilization close to the desired 100%. That didn’t mean that 100 licenses suddenly became sufficient to get the jobs done. Rather, the user had a scalable methodology to fully utilize all the licenses that he wanted to buy. Today, that same user is running close to 10,000 concurrent tools on a large compute farm.
While the tool and job scheduler vendors didn’t work together directly, they formed a loose association to better serve the customer and helped themselves as well. The customer purchased many more tool flows based on business and time-to-market requirements.
If only the system design ecosystem could apply a bit more of a customer-centric view and apply it more often. In this example, the user community sits at the top of the ecosystem and tries to maximize the amount of processing throughput that it can get for the lowest amount of investment. Vendors concerned that tools such as job schedulers cause lost revenue could think in terms of eliminating constraints instead and work cooperatively to offer attractive licensing packages as a way to extend an existing system’s capabilities.
Mindfully taking the attitude that “the customer is always right” could have a lasting and positive economic impact. Boosting customers’ efficient use of their resources will accelerate growth, not impede it.
— Andrea Casotto is founder and chief executive officer of Runtime, developer of software products for the efficient management of computing resources used in the design of complex chips, including processors and software licenses.