“The next time I get an accurate forecast from a customer will be the first time I get an accurate forecast from a customer.” – Common supply chain gripe
Customers are the supply chain’s raison d’etre, but they are also its biggest challenge. Customer forecasts are notoriously inaccurate. In the past 20 years, that’s twice had a devastating effect on the electronics component ecosystem.
In 1999, unbridled optimism about the dotcom revolution fueled excessive component demand that was followed by a costly — $13 billion — inventory glut. In 2020, chip demand from carmakers unexpectedly surged, triggering a global IC shortage that’s entering its third year. The auto industry has lost $210 billion worth of production.
Source: Flex
Efforts to capture a global demand picture have been elusive because forecast data still resides in silos within OEMs, EMS providers, distributors and component suppliers. Items such as customer identity, contract pricing and order volumes are considered proprietary information within the electronics supply chain. But an industry-wide single version of the truth, according to one top electronics executive, would contribute toward better decision-making within this complex and interdependent ecosystem.
A project called True Demand was outlined by Flex’s Chief Procurement and Supply Chain Officer Lynn Torrel at the ECIA’s October Executive Conference. The Covid-19 pandemic and an unprecedented scarcity of chips over the past two years is prompting a reexamination of how the supply chain historically has worked. Shortages are usually followed by a buildup in capacity, excess inventory and price erosion.
“Our vision statement is rethinking the ways we forecast demand and communicate information and how we can collaboratively develop solutions that mitigate supply chain disruptions and temper future bullwhip effects,” Torrel said. “This isn’t a silver bullet. It’s not the only thing that needs to be done, but it’s a step in the right direction.”
True Demand would use an independent third-party (ITP) to collect forecast data from industry stakeholders and compare it with industry growth rates and other macro indicators. Company information would only be used for aggregating data and participants’ identities would remain private. The ITP would then make recommendations to aid in decisions such as capacity planning and investment. Industry-wide participation is critical to the effort.
The bullwhip effect
Most chip makers directly manage only a select group of their largest or most strategic customers. The remainder of their forecasts come from distributors, EMS providers and other partners. During component shortages customers have been known to place the same order with multiple suppliers, falsely inflating demand signals. Following the inventory glut in the early 2000s, multiple chip foundries shut their doors. Since 2020, plans to build more than 30 fabs have been announced.
“I love my customers, but forecasting is not good in our industry,” explained Torrel. When component lead times extended, Flex’s customers front-loaded their demand anticipating they’d get half of what they needed. “The messaging I heard from customers and our business units was ‘we’ll take whatever we can as soon as you can manufacture.’ So just get that product and we’ll take it,” said Torrel. “I could just see that we were creating the same situation we had in the past. Right? Demand outstrips supply which leads to supply chain challenges and significant investments are made to increase supply.”
Chip orders began accelerating early in 2020 and by Q3 of 2022 fabs were seeing orders pushed out or cancelled. Inventory is building and consumption is weakening, according to research firm TrendForce.
Source: Flex
Torrel, who also worked at global distributor Avnet Inc., consulted peers, academics, market research firms, industry trade associations, component makers, distributors and other stakeholders in the electronics supply chain. Unsurprisingly, many companies asked: “How does [True Demand] benefit me?” Electronics distributors, in part, exist to manage customers’ upside and downside forecasts. One ECIA attendee expressed concern regarding the future of the distribution industry; other companies have declined to comment citing their business relationships with Flex.
Augmenting supply chain data
“True Demand’s model is to augment, rather than replace, the supply chain ecosystem,” Torrel explained. During the shortage companies were focused on their own supplies and operations. But double ordering, the open market and other practices to secure supply impact the entire industry.
“Over the past year, if think about the broker base, exorbitant prices are being charged by brokers and the reality is that they’re still getting their hands on the parts,” she said. “We need that inventory to keep our lines running.” Orders have constantly been expedited, which adds to costs.
Through True Demand, companies would input data that’s protected by the ITP. Macro information – such as industries that are growing or contracting, would also be analyzed. Torrel used the example of smart meters, an industry that’s forecast to grow by 20 percent. The four biggest players in that market use that 20 percent to estimate growth for their companies. Their demand is based on that forecast.
“Historically that’s been acceptable,” Torrel noted. “We know those demand signals are reverberating in the system. The vision of True Demand is it would normalize that. It would be able to identify the type of industry and consolidated industry growth.”
The True Demand project reached out to Microsoft and found its Azure platform could manage the process and keep data secure. “We’d all continue to forecast demand like we do today, the OEM to the EMS, the EMS to the distributors or to suppliers, and suppliers would do their modeling,” said Torrel. “But in addition, we’d send those signals into the independent third party, and again, you have each part number; an OEM part number tied to a manufacturer; or a part number by product — however [companies] do the analysis.” The system would incorporate AI and machine learning.
Plans include forming a 501(c)(6) – a non-profit coalition – prior to moving to proof-of-concept. Torrel emphasized the importance of industry-wide collaboration.
“I think it’s important to remember as supply chain professionals living and working through this incredibly disruptive environment, we have an obligation to learn from each other and to look for ways in which we can work together collaboratively to drive a better future,” she concluded.
