In the digital economy, electronics are once again resembling the old “heavy” industries: without available megawatts, critical materials, or advanced manufacturing capacity, growth becomes a difficult-to-realize promise. This is the picture emerging in 2026, with Artificial Intelligence acting as an accelerant for trends that have been brewing for a long time. Underlying this is a concept that discomforts governments and companies alike: the bottleneck is no longer just about designing better chips, but about powering, building, and assembling them at scale.
The Global Electronics Association (the leading international industry organization, legally known as IPC International Inc.) has been emphasizing a concept for weeks: the map will shift toward strategic diversification, not toward a “total disconnection,” which remains practically unviable in a globalized supply chain. Simultaneously, the geopolitical tension over raw materials and critical technologies is escalating. The result is an industrial planning scenario increasingly resembling risk management.
Electricity sets a limit on AI expansion
The boom in accelerated computing has turned data centers into a matter of national infrastructure: electrical grids, substations, permits, cooling water, and supply stability. It’s no coincidence that recent reports no longer ask “How many GPUs fit?” but rather “How many megawatts can be contracted, and within what timeframe?”
The International Energy Agency (IEA) estimates data centers’ electricity consumption will be around 1.5% of the global total in 2024, with strong growth projected over the next decade driven by AI and accelerated computing. In Europe, the European Commission summarizes this trend with a figure increasingly cited in public reports: the energy demand associated with data centers could more than double by 2030, with AI as the main engine.
Industry-wise, this translates into three very specific movements:
- Location selection based on energy availability, not just connectivity: where there is real and predictable electrical capacity.
- Cooling as a competitive advantage: from air optimization to the growing adoption of liquid cooling and immersion systems in high-density projects.
- Redesign of the power “stack”: from power electronics to in-data center distribution, with emphasis on availability and delivery timelines.
Rare earths: from invisible input to strategic asset
If energy is the visible “brake,” rare earths and critical minerals are the silent factor shaping technological sovereignty. Neodymium, praseodymium, dysprosium, and their companions support permanent magnets for motors and generators, but also impact power electronics, robotics, and an increasingly large portion of the industrial equipment surrounding AI.
The Global Electronics Association anticipates that by 2026 raw materials will be treated as matters of national security, prompting more aggressive investments in mining and domestic processing—even when economic conditions are less favorable. The inconvenient truth is the concentration: analyses like those from the Center for Strategic and International Studies (CSIS) highlight China’s dominance in much of the ecosystem, with very high quotas in separation/processing and magnet manufacturing, as well as a decisive role in mining.
Operationally, companies are preparing for a world where “just-in-time” is losing its appeal:
- More critical inventory and dual sourcing, even if more expensive.
- Regionalized supply chains (not closed): redundancy and alternative routes.
- Greater traceability and compliance, as the origin of materials becomes as relevant as their price.
Packaging is no longer a detail: it’s an industrial battleground
For years, the headlines were about nanometers. Now, the conversation increasingly shifts to how complex chips are assembled: chiplets, stacked HBM memories, interposers, 2.5D/3D packaging… In other words, advanced packaging.
The reason is straightforward: to scale performance per watt and bandwidth, a “monolithic chip” is no longer enough. The value moves into the entire system within the package, and here two realities emerge: industrial capacity is limited, and the technologies are extremely demanding.
In this context, solutions like glass substrates are beginning to gain traction as responses to the physical limits of traditional organic substrates: flatness, mechanical stability, thermal expansion, and finer design rules. Intel, for example, advocates that glass could open new possibilities in package size and interconnection density. Simultaneously, technologies like hybrid bonding are consolidating as key components of high-density 3D integration, though recent technical debates highlight clear challenges in defectivity, metrology, and cost within the packaging ecosystem.
The macro impact is direct: even with new factories, the industry may face “bottlenecks” in advanced assembly. Financial media have pointed out that capacity shortages in advanced packaging (such as CoWoS and similar) are becoming factors that influence schedules and ramp-ups in AI and HPC.
What does 2026 mean? Engineering under pressure, procurement within design
The most significant consequence isn’t technological but organizational: design and procurement are becoming inseparable. Choosing a component is no longer just based on datasheets but also on its lifecycle, multiregional availability, ease of substitution without redesign, and “country risk.”
Therefore, 2026 is shaping up as the year many organizations formalize practices that were once merely “recommendations”:
- Lifecycle visibility: avoiding costly redesigns due to obsolescence and supplier shifts.
- Replacement-tolerant architectures: interchangeable components with minimal impact on certification.
- Geopolitical risk management: from copper and rare earths to packaging materials and advanced memory.
- Energy planning: not just as an add-on, but as a design requirement for products and data centers.
The sector isn’t “stopped”; it is re-prioritizing. The clear takeaway: the next competitive advantage won’t just be inventing faster but building reliably in a world with limited energy, politicized materials, and intense advanced assembly pressures.
Frequently Asked Questions
Why can a lack of electricity slow down AI data center projects?
Because modern AI increases power density per room and requires high electrical contracts, substations, and connection timelines that don’t always keep up with deployment pace.
What are “rare earths,” and why do they affect technological sovereignty?
They are a group of key elements for permanent magnets and industrial components. The concentration of processing and magnet manufacturing in few regions makes their supply a strategic factor.
What is “advanced packaging,” and why is there so much talk about CoWoS, chiplets, and HBM?
It refers to technologies that integrate multiple chips (compute, memory, I/O) into a single high-interconnection package. In AI and HPC, this assembly is crucial for performance and efficiency.
What steps can companies take to mitigate risks in the electronics supply chain by 2026?
Plan components early, demand lifecycle visibility, design for substitution, diversify suppliers and regions, and treat energy and logistics as engineering variables, not just procurement issues.