CHIPS and economic dips: Four ways AI and cloud help chipmakers thrive in a volatile world
Matt A.V. Chaban
The semiconductor industry was grappling with fragile supplies even before the pandemic upended almost everything. New tech can help cope.
It was a big week for the semiconductor industry in what’s already proving to be a big decade.
On Tuesday, U.S. president Joe Biden signed the CHIPS and Science Act into law. Versions of the bill had been circulating in Congress for years, with the one just approved first introduced in June 2020. While that was months before the semiconductor shortage truly took hold, backers of the bill knew disruption was coming and seized on the COVID-19 crisis to advance a strategic imperative. For many nations, the $556 billion semiconductor market isn’t simply a means to create well-paying manufacturing jobs; it's also where the building blocks for data centers, devices, even dog collars come from.
And yet just as the past two years have shown how essential the world’s fourth most traded good is, public investments from Washington—as well as Amsterdam, Beijing, Taiwan, and Tokyo—still won’t be enough to insulate the industry from the ongoing disruptions that got us into this microchip mess. On Wednesday and Thursday, some of the top chipmakers released challenging forecasts. Fueled in large part by a sudden overabundance of their wares, electronics companies are feeling the whiplash of an unevenly cooling economy.
“In the conversations I’ve been having, many executives recognize they can’t change the situation we’re in—no amount of stimulus can overcome volatility—and they’re just looking to manage it better,” Simon Floyd, industry director for manufacturing and transportation at Google Cloud, said in an interview. “They’re looking to focus on making the right things in the right quantities, getting inventories as just-right as they can. And then amidst all that, they’re working on their long-term plans to tap into the opportunity that comes from CHIPS and other investments, which really are going to transform the industry over the next decade.”
Many, it turns out, are turning to cloud- and AI-based technologies to help them navigate both the short- and long-term challenges and opportunities before them.
Floyd, and his colleague Peeyush Tugnawat, a principal architect for electronics at Google Cloud, have been developing tools for the industry for decades. When the CHIPS Act passed, we reached out to them to find out where chipmakers should turn their attention and resources. They outlined four key areas where the cloud is already having an impact:
High-performance computing for chip design
Supply chain intelligence for components and materials
Manufacturing operations optimization
Logistics and sales to downstream customers
“Right now, companies are looking for two things: they need to keep innovating, and they need to maintain business continuity,” Tugnawat said. “They’re finding opportunities throughout their operations to enhance both with the cloud.”
Strengthening supply chain intelligence
Just as countless companies are waiting on semiconductors these days, chipmakers are equally desperate for the resources for their products. The collection, refining, and shipping of raw materials continue to be interrupted around the globe. One of the most compelling ways to mitigate this is through greater visibility vis-a-vis a digital twin of your supply chain.
A dynamic virtual representation of each step and component of supplies, a digital twin, when fully deployed, has the capability to zoom in on a single shipment or component for pinpoint accuracy; it can even be connected to secondary and tertiary suppliers for added visibility and predictability. When overlayed with AI, the digital twin gains the ability not only to track but draw insights over time from supply flows, which can enhance predictions and decision making. AI can also be used, either within the digital twin or in a separate dashboard, to analyze external events, such as financial trends, natural disasters, or geopolitical events.
And while not directly related to supply chain resilience, digital twins have the added benefit of providing more precise and transparent sourcing information. This can help when it comes to ensuring materials were collected sustainably, ethically, and honestly, which both consumers and regulators are increasingly demanding.
Optimizing manufacturing operations with data and AI
The CHIPS Act is only one of many public investments nations are making in driving new chip production—hedges against not only uncertain supplies but also political instability. By some estimates, China has spent more than $100 billion to boost semiconductor development the past two decades, twice the amount in the CHIPS Act.
Whether such investments occur in new, old, or refurbished facilities, every worker, every component, every machine, and every product is a potential point of data that can amplify these critical expenditures. Optimizing in each setting presents its own challenges and benefits—beyond the amortization costs, updating legacy environments may not be as hard as it seems, given how cloud and edge technologies are easily integrated by design.
A platform like Google Cloud’s Manufacturing Data Engine can help thread together all the different aspects of the manufacturing process. Production managers gain rich, real-time insights into the depths of their machines and predict future events or conditions that can be mitigated. Operations leaders can monitor multiple systems remotely. AI that learns and improves over time can not only catch but potentially preempt mistakes and drive new levels of optimization. Especially as a labor shortage for qualified factory workers continues, the ability to automate and support more processes with limited workers becomes essential.
AI for quality control is another area of particular interest for manufacturers (so much so, we just published an entire article about quality inspection tech), with particular benefits for chipmakers. Their products are so small and sensitive, with incredibly tight tolerances, that it can be especially challenging for human inspectors to pick up on defects. And with the high cost of semiconductors, catching defects faster, and preventing them in the first place, becomes all the more important. When some companies are spending as much as 40% of revenues on quality-related issues, catching them faster, or eliminating them altogether, pays serious dividends.
High-performance computing to accelerate chip development
Cloud is also driving innovation beyond production, with notable impacts on design, as well. Developing and testing the layout of semiconductors is becoming increasingly challenging, especially as they become more specialized, more miniaturized, and more energy efficient. Whether fabless or integrated, both operations are looking to speed up the testing of designs to iterate on products and accelerate time to market.
High-performance computing has become a popular tool for chip design because it accelerates modeling and testing of designs before putting them into production. Previously, this would have meant high-powered machines on-site, an expensive proposition. Now, almost any organization can access high-powered computing resources through the cloud, leveraging them to run their models without the upfront and quickly outdated hardware costs.
There’s also a potentially crucial integration here with the supply chain solutions. Chipmakers have to be vigilant that the component they need for their designs will be available—designing to their supply, in other words. With supplies changing so quickly, high-performance computing can help with iterating on designs with the most up-to-date sourcing. And should your supplies change unexpectedly once production is underway, the faster computing resource can help companies more quickly recover by being able to redesign in a timely fashion.
Chipmakers are focused on making the right things in the right quantities, getting inventories as just-right as they can. And then amidst all that, they’re working on their long-term plans to tap into the opportunity that comes from CHIPS.
Gaining insights from new sales and logistics channels
One thing manufacturers have often lacked—and one of the biggest changes coming about through the cloud era—is the ability to follow their products to customers and end-users, and all the insights that can come with those connections.
New data analytics tools can help develop a deeper understanding about partners, customers, and consumers, the goods they are seeking, and how they are using them. At a moment of volatility, this can be crucial information to adapt your supplies and products. There are also a range of tools around asset tracking and last-mile solutions that help monitor timely deliveries, which help keep eager customers satisfied amidst chaos and competition.
Another area manufacturers within and beyond the chip space are looking is to create more user-friendly and direct-to-consumer-style sales portals. These tend to mimic the experiences consumers have come to expect from non-industrial brands. People want to buy things the easy ways they’ve grown accustomed to, whether it’s soap, sushi, or semiconductors. These portals can help boost sales, and they contain yet more actionable data the company would not otherwise have if it was engaged in more traditional distribution models.
One moment, chips are in demand everywhere. The next, there's an oversupply. Manufacturers are looking for better insights to manage their supplies and products.
Chips start with stronger workers
Between the opportunities of CHIPS and other public investments, and the challenging economic headwinds ahead, Floyd and Tugnawat said electronics companies are going to have to find even greater focus than they have the past two grueling years. It's nonetheless an exciting moment, given historic shifts on the horizon, and the reality that semiconductors will remain an increasingly important component of almost every business going forward.
When it comes to prioritizing, chipmakers can begin by focusing on their workforce and how they want to augment them with cloud technology. If the objective is increasing productivity, because organizations are resource constrained, AI can boost workers on the factory floor, helping them do more regardless of supply and economic circumstance. If the goal is upskilling workers in order to retrain them, AI can help by removing repetitive, low-value tasks. If the aim is creating more innovative products with more constrained supplies, workers can level up through high-performance computing, creating immediate returns.
"What I love about CHIPS is, it’s all about being a world leader in electronics," Floyd said. "And everyone will want to be that leader now, regardless of company or country."