We hear a lot about outsourced manufacturing these days. The complexity it layers onto a supply chain is well acknowledged. But what are some of the issues that arise and how can you solve them? This month, we take a look at the business pains of two well-known companies in the electronics industry—an industry where supply chain volatility already is a given. These issues are complicated by shortened product lifecycles, constant revamping of supply chains and the need for absolute product perfection and on-time delivery. We’ll see how one Silicon Valley solutions provider helped stop their pain.
Let go of the spreadsheet
Juniper Networks is a $4.5 billion provider of high-performance network infrastructure, headquartered in Sunnyvale, Calif. The company designs and sells software, silicon and systems that transform the experience and economics of networking. Its broad product portfolio is made by 10 manufacturing partners—four primary—in China, Malaysia, Taiwan, Mexico and the U.S., meaning visibility becomes vital.
“A couple of years ago we had a tight supply environment,” explained Craig Martin, Vice President of Operations, Juniper Networks. “A lot of our suppliers were extending lead times and de-committing. It was pretty tumultuous.” Given that Juniper outsources all its manufacturing, it does not have classical Manufacturing Resource Planning (MRP) and instead relies on its manufacturing partners to order and manage materials.
Juniper tried to track inventory shortages by having each site do an update and use Excel. That didn’t work too well, said Martin.
“Because it’s a combination of time latency and the static nature of Excel it’s only good until the moment you put it in the spreadsheet,” he explained. “It becomes outdated right after that. It was difficult keeping track of where we were given multiple daily changes we were getting. You’d come in, in the morning and thought you had a picture of what your supply was—but by the afternoon it was changed. The challenge was not having real time. It would be suspect as to whether it was really legitimate because of the time latency.”
Components with a normal lead time of eight weeks were changing to 18 or 24. Juniper executives decided that they needed a solution that would give them real-time views of what was on hand and on order at each site. If there was an excess at one site and a shortage at another, a phone call could easily make a change.
After a several months-long search, Juniper Networks chose Serus Corp., a Software as a Service (SaaS)-based provider of technology and managed services for companies with distributed or outsourced manufacturing operations.
“With the Serus model, they understood what we were looking for and how to go about getting it to us in a standard way without a lot of customization,” said Martin. “It was a much more compatible match in terms of how we wanted to go about things—a very straight-forward process.”
“Serus gave us the tools to tap into each of our manufacturing partners’ databases for real-time visibility from their MRPs,” he continued. “The situation was so dynamic. We were able to leverage our external partners. We saw their prioritization, their internal structure and resources and got details when red flags turned up. A lot of dollars are flowing through our partners to support our business and it demands the kind of intelligence and discipline we get from the Serus tools,” said Martin.
Get it to market
Santa Clara, Calif.-based NVIDIA Corp., a visual computing technology provider, ships anywhere from 25 million to 35 million chips each quarter. Because the company ships that many chips to so many different market segments and customers, its’ very complex supply chain undergoes constant change.
A few years ago, the company realized it had two problems to solve. One was increased visibility into that complex, ever-changing supply chain. The other was how to handle the short product lifecycle of the chips.
“We needed systems agile enough to basically change the needs and requirements we had for fitting the products out,” confirmed Sunny Bedi, Senior Director, NPI & Program Management, Operations, NVIDIA. “We had a huge amount of touch time for the product engineering team or test engineering team to have to go into the system and update yields or test programs or any other relative data that is important for our factories to know.”
“As we are completely fabless and completely outsource manufacturing, it’s important that the information our factories have is accurate—or else we basically build the wrong part and it goes out to the customer,” Bedi added.
NVIDIA previously used several costly custom applications. The company wanted to simplify and streamline operations to maintain and update the supply chain much faster and with the agility necessary for the product lifecycle, which can be as little as six months.
It chose Serus to improve product engineering productivity, reduce manufacturing errors, increase supply chain velocity, improve chip quality and improve projected manufacturing metrics to improve decision making.
“If I had to summarize, I’d say the benefits for us are increased engineering productivity,” said Bedi. “We used to have a lot of manufacturing errors on the floor. We’d get a call in the middle of the night—China or Taiwan time—and the engineer would be on the phone fixing it. We don’t want product engineers trying to solve system or data problems. It’s a poor use of their time.”
In the end, Nvidia’s product engineering productivity improved by 40 percent; manufacturing errors were reduced to zero; cycle time for assembly and test improved 15 to 20 percent; and chip quality was improved due to decreased manufacturing deviations.
“If your data is accurate, your information is accurate,” said Bedi. “And if your information is accurate you can make better decisions. Data accuracy is so critical and all of that has been significantly improved by this solution,” he concluded.