The global electronics supply chain stands at a precarious crossroads. For Andrea Klein, CEO of Rand Technology, the current trade and tariff environment – coupled with seismic shifts in technology adoption – has created one of the most complex and uncertain periods the industry has experienced in decades. But also, one of the most exciting.
In this exclusive Q&A with Procurement Pro, Klein reflects on tariffs, trade, AI, and the road ahead for the electronics industry.
How would you describe the current global trade and tariff landscape, and what do you see as the most significant implications for the electronics industry and its supply chains?
There was a lot to unpack in that question, but to summarise on a macro level: the world has spent the last 50 years globalising, and to now expect a complete shift to regionalisation is unrealistic. Whether it is raw materials, minerals, or components, the idea of establishing full component manufacturing capacity in every region is not only impractical, but financially disastrous. It is simply not cost effective, and expecting companies to regionalise across the board makes little sense.
That said, the current environment – particularly the reintroduction of tariffs – presents its own risks. Historically, tariffs have had a devastating effect on economies. In fact, one of the true causes of the Great Depression in the United States was tariffs. So, in summary, regionalisation is not viable in the near term, and there will have to be some kind of compromise.
Looking ahead, the industry needs to consider the technological shift we are undergoing. Between 1990 and 2020, the market was defined by compute and mobile technologies, driven by the Internet. That cycle has ended. Since 2020, we have entered a new cycle driven by artificial intelligence. We are now in a technology revolution where hardware has re-emerged as critically important.
Over the past three decades, hardware became a mature industry with declining margins. Around 10 years ago, executives across the hardware sector recognised that they were no longer generating sufficient profit. Many shifted their focus to software, where margins were higher. Hardware began to be viewed as a financial burden, leading to increased outsourcing and cuts in commodity management and supply chain investment. This wasn’t necessarily about disrespect for hardware functions, but rather a psychological devaluation of their role.
As a result, the sector stopped developing the next generation of hardware talent. There was little to inspire new entrants into the field, especially across Western markets. By the time AI took centre stage post-2020, we found ourselves facing a talent gap. Baby boomers had retired during COVID-19, Generation X is relatively small, and Millennials show limited interest in this area. While Asia and India continue to develop talent, Western countries must now catch up.
AI adoption is still in its infancy. It is surprising to many that ChatGPT only launched around 18 months ago. Though AI has existed for years, its current adoption and practical use cases are still emerging. Right now, it is expensive and complex. Infrastructure is being built – especially data centres – which require power, thermal management, compute, storage, networking, and connectivity. Eventually, as tools and applications mature, AI will shift to the Edge. This will drive massive demand for hardware, memory, power, and thermal solutions.
In terms of market dynamics, 2021 and 2022 experienced a demand-driven shortage caused by the post-COVID rebound. Demand surged across all sectors, and suppliers operated at full capacity. To manage this, suppliers introduced long-term agreements with extended lead times. The idea was to identify true demand – customers would hesitate to commit unless orders were real. But the opposite happened. Everyone signed the agreements, fearing they would be left short.
The correction came in 2023 and 2024. Demand dropped sharply, and only the AI ecosystem continued to grow – accounting for 15-20% of the market. The remaining 80% effectively stalled. However, the contracts remained in place, and inventory kept arriving. The result was a significant oversupply. Companies had to manage inventory, cash flow, and credit liabilities. Revenue fell, and suppliers could not raise prices due to weak demand. Their only option was to cut investment and utilisation – down to 50-60% in many cases.
Some suppliers had invested in additional capacity during the earlier boom and weathered the correction more easily. But heading into 2025 and 2026, only AI continues to provide significant momentum. Early in Q1 2025, there were signs of stabilisation and hope for recovery in the second half of the year. Then came the return of tariffs, creating fresh uncertainty.
Outside of AI, customers remained cautious. Those with surplus inventory and low demand were driven by CFOs to under-forecast and avoid further stockpiles. As a result, visibility into demand remains poor. Suppliers, in turn, are reluctant to invest in additional capacity. This brings risk: we could see a global recession.
Q2 2025 is proving pivotal. Most sectors – AI, automotive, industrial – have finally normalised inventory after more than two years. However, because they have under-forecasted for so long, even small increases in demand are causing disruptions. Lead times are extending to 26 weeks or more. For the first time in over two years, parts are no longer available within weeks.
This marks the beginning of a supply-driven shortage – the first since 2008-09. Suppliers cut investment, customers cut forecasts, and now upsides are returning. Even in quiet sectors, shortages are beginning to appear. Prices are starting to rise, despite the fact that demand forecasts remain uncertain.
The longer-term issue is that once AI applications reach the Edge, demand will grow exponentially. But the supply capacity needed to support that growth is not yet in place. Component suppliers who had cut utilisation are now seeing lead times grow: from 12 to 15, then 20, 26, and now 36 weeks. Some are moving to allocations. Passive component makers, who did not invest during the boom, are now operating at 90-100% utilisation. That is a dangerous position to be in.
Supply chain challenges are set to intensify. Shortages and price increases should be expected. Geopolitical tensions add further cost and risk. Companies must now maintain inventory across multiple regions, unsure where their next safe harbour for production will be. In the short term, the best risk mitigation may simply be to bring in inventory during a grace period and hold it.
This is a highly delicate moment for the global electronics supply chain. Companies not involved in the fast-moving AI ecosystem must stay alert. They may not be the ones driving the shortages, but they will almost certainly suffer the consequences.
What would your what would your advice be to help customers mitigate these challenges over the next three to four years? And how is Rand positioning itself to support customers through these challenges?
I would simplify the BOM. I’d be very careful about sole-source products. I’d aim to ensure that at least 95% of the BOM had second-source options available. It’s expensive, but I’d also build out models that allow for the use of different products on the same board – essentially, two different BOMs for the same product – to create flexibility.
I’d also get out to the suppliers and focus on building in as much flexibility as possible into my boards. Let software dictate where it can, but don’t let it be the only driver.
Rand has been in business for 33 years and we’ve built up a tremendous amount of data. We’re well known for our market information and predictability. What we’re doing on behalf of customers is creating reports that track and provide heat maps – not just for lead times and pricing, but also activity levels in the market.
We’ve been able to supply enough information to guide our customers in buffering their inventories and providing alternative solutions for manufacturers globally. We’ve become a significant source of both information and guidance, helping customers develop solutions to their problems.
How do the problems that we’re facing right now compare to problems that that we’ve faced before (Recession, COVID-19 etc.)?
Well, that’s a great question, Paige, because a lot of history repeats itself. Customers tend to do the same thing, but this time is different. I don’t think people fully understand the magnitude of what’s coming with AI and the impact it will have on hardware.
Everyone has become accustomed to the last cycle, where you’d build a box, then make it a bit faster or smaller. But now, we’re seeing entirely new technology with very different types of hardware. This explosion of AI is a very exciting time, but I don’t think people are anticipating how fast it will happen, how large the impact will be, or how unprepared we are.
If I had to offer an analogy, I’d compare it to the late 1980s, when Windows came out and the PC explosion took place around 1990. The adoption of computers and electronics was so new that it took people’s breath away. I believe we’re heading into something similar again.
The misconception right now is that AI is all about the software – Gemini, ChatGPT, and so on – but people don’t realise the extent to which hardware will need to change. This will significantly affect components and the supply chain, and I don’t think that’s being anticipated at all.
Think about how much more memory will be required. Imagine a shift where everything is GPU-based rather than CPU-based. This will be exponential in scale – exciting, yes, but also difficult to manage.
Do you think that excitement and explosion in AI innovation is going to help attract the younger generation to the industry?
It will attract people to the software side of things, but not to hardware – unless we do something fundamentally different. I don’t believe that Millennials are drawn to the heavy lifting involved in a hardware supply chain. There’s a lot of complexity and manual effort involved, and we haven’t really tried to make it exciting for them. It’s not the kind of workload they’re typically interested in.
You can’t fully automate it. Even if you use AI to support your supply chain, it remains an imperfect world with many variables. That part is going to be extremely challenging, and I’m not seeing enough mentorship happening to address it.
Unfortunately, with this next difficult cycle in the market, I think the older Gen X professionals are going to start leaving too – much like the Boomers did during COVID. So, we need to treat this as a serious issue: how do we get people trained up? I think there’s potential with Gen Z – they seem to genuinely want to help, and I see a real opportunity there. As for Millennials, perhaps there’s interest on the software side, but I just don’t see them moving into hardware.
