As we referenced in our previous blog, three major trends in the form of de-globalization, economic nationalism and a return to the era of Great Power politics, are currently reshaping the world. Perhaps, most especially, the semiconductor world. These forces both represent significant challenges to negotiate, but equally, opportunities to exploit. As we will explain below, this is particularly true for Start-ups and SMEs. For larger players, however, tariffs and export restrictions could create a chasm between winners and losers.
Figure 1. AI and Non-AI Chip Actual Revenue 2024 and 2028 Forecasts.
Before delving into these impacts in more detail, we need to re-state the context. In reviewing defining global trends since 2019, it can be strongly argued that you cannot fully understand the past six years without putting the semiconductors at the center of consideration, and you cannot understand that industry without you placing AI —including quantum applications—at the heart of the concluding analysis. With this in mind, the final and definitive context must, therefore, be that of the advanced package architectures needed to produce these chips. By “advanced” we mean a range of packaging technologies that enable reduced pitch, chiplet interconnectivity, and increasing numbers of actively aligned photonic connections; at a bare minimum this means utilizing bumps of ≤100µm, and that the direction of travel is moving towards the 50 to 5µm range over the medium-term. The latter will define the parameters for next generation 2.1-2.5D architectures, including micro-bump TCB/flip-chip processes. Hybrid and 3D and 3.5D architectures are, and will increasingly be, even more technically demanding. Perhaps requiring a pitch range that over the next three to five years will work within a tolerance of 2-3µm and finally 1µm by the early part of the next decade. Before then, however, 2026 will witness production of the first chips embracing 3.5D architectures. These allow still greater density and higher performance so foundational to the future of advanced AI solutions. Collectively, these remarkable steps represent the enablers of sustainability and growth across the entire microelectronics industry.
As we previously acknowledged, demand for non-AI chips, even after a period of seemingly indeterminable downturn, remains decidedly lackluster, with growth—by which we mean revenue growth— limited to the low, single-digit range. Indeed, much of the data from this year so far throws even this into doubt.
Figure 2. Palantir Stock Value October 2020 to Mid-May 2025.
Even longer-term forecasts predict sluggish performance for non-AI chips as we move towards the close of this decade, assuming the global drives towards net-zero remains on its current trajectory. At least in the West, economic growth can no longer be taken for granted. Governments are increasingly realizing that net-zero and increased domestic consumption are proving somewhat incompatible. Add to this the impact of tariffs and headwinds becomes the default position. Beyond this, even if growth were to take place price pressures mean non-AI chips will only remain profitable if produced at ever greater levels of efficiency.In contrast, and on the basis that SME business success is predicated on “following the money” we direct the reader’s attention of Figures 1 and 2. Reviewing historical media coverage makes it clear that the current dominance of, and projected future percentages allocated to, the size of revenue from AI, including its quantum aspects, have largely surprised ecosystem analysts. Not least those market leaders whose development strategy was predicated on this shift. That being acknowledged, it is not the fact of this shift, but the degree of this shift that has proven largely unseen. There can be no hiding the huge impact that success in the sphere of AI has on stock value, commercial success and further investment in research and development. The dramatic impact of AI is little better illustrated than in the fact that seemingly out of nowhere, Palantir has become the US largest defense company by several times over.
Figure 3. Advanced and Hybrid Packaging Processes and Clean Room Requirements.
While there is general consensus among commentators that levels of investment into the AI sector, and not least specialist start-ups and SMEs, is running at exceptionally high levels, current stock values of the 10 largest other public players makes explicit the same story. Of course, stock prices are only indicative of future expectations (and the pursuit of investor profit) that being said, these expectations are based on a usually thorough review of a significant range of data including government policy, consumer demand and technical trends.
Figure 4. Simplified Overview of Advanced Packaging by Density and Performance.
If Figures 1 and 2 show that technical development and investment inflows are converging on AI, the third and definitive factor must be the processes by which these AI and quantum related devices are manufactured. Most fall entirely in the range of the advanced packaging technologies illustrated in Figures 3 and 4. In reviewing system requirements for micro-bump, thermo-compression, and flip-chip capabilities, these require very tight accuracy parameters, often >1 micron, a fact made obvious by the clean room requirements set out in Figure 4. This may be a simplified overview, but the direction of technical travel and investment that must be pursued by SMEs who want to participate—or more correctly specialize—in these heterogenous packaging opportunities, is unambiguous. Equally, there is a need to understand the implication of the fact that these processes require a complete rethinking of chip design and prototyping, especially in respect to chiplet and hybrid approaches.
SMEs must realign their thinking accordingly, perhaps radically. Thus, the lessons for these companies are not only those of aggressive learning, aggressive customer engagement, technical, commercial and business agility, but of foresight in investing in the right capabilities and ensuring a solid bank of people are trained to operate them. The latter can no longer be taken for granted. This is true even of ecosystem giants such as TSMC whose $165 billion US expansion plan has been in part slowed down by the difficulty in finding the right human resources.
One more thing is needed: capital. Investment always makes SMEs nervous, or at least CFOs, but the key here is, as we have implied, of leach-like attachment to the customer and participation in local clusters for co-development opportunities, but especially leach-like attachment to the customer. Beyond this, what is required for successful engagement in developing a product using advanced packaging is scientifically grounded self-belief together with unswerving focus.
The forces of de-globalization, economic nationalism and Great Power hegemonies, mark levels of uncertainly unknown since the end of the cold war, a potential rise in inflation and supply chain disruptions brought on by tariffs, increased nervousness in discretionary spending brought on by ideological conflict, and potential military stand-offs. Mindful SMEs with see through these elements as surmountable challenges to be unemotionally negotiated. In this manner headwinds will definitively morph into a new epoch of opportunity. Not least because winning the AI race—to quote Congress “the second Manhattan project”—is heavily predicated on start-ups and specialist SMEs. SMEs are recognized as essential to our collective success. The range of recent acquisitions by market leaders underscores this. We need to learn from the fact that the SMEs so acquired have a singular, distinct and very defined place in the continuum that is concept to final product. Equally, by definition these are high margin opportunities for those able to match the underlying value proposition.
In terms of the broader context set out in our first blog, the Trump Administration is adopting policies that will affect the long-term restructuring of the global higher technology manufacturing landscape. Semiconductors are now recognized as strategic assets that must be protected through resilient local supply chains.
For these reasons, Palomar Technologies is investing heavily in our newly created Advanced Solutions Division, formerly known as Assembly Services, both in the US and in Singapore. As we roll out new capabilities to advance the development and testing of cost effective heterogeneous integrated architectures, we will be informing current customers and other interested parties. Our goal is to support our customers in the discreet—and often hybrid processes—that take their product into future markets. With this in mind we invite our readers to make known to us your advanced packaging needs. We always make time to talk.
Our final blog in this series will examine the potential impact of commercial and geo-political trends on the Defense sector.
----
Dr. Anthony O'Sullivan
Strategic Market Research Specialist
Palomar Technologies, Inc.