The semiconductor industry has officially entered the “2-nanometer era”. And, as happens every time the sector moves to a new technological level, the news isn’t just about smaller transistors: it’s about who can manufacture on time, at what cost, and for whom.
In this landscape, TSMC — the global leader in contract manufacturing (foundry) — has confirmed that mass production of 2 nm has begun as planned, during the fourth quarter. The issue for much of the ecosystem isn’t whether the technology works, but that the demand from major clients is straining the supply chain to a point that forces looking for alternatives… and that’s where Samsung Foundry emerges as a more serious option than many expected at this stage.
The bottleneck is no longer just technology: it’s capacity
According to the South Korean newspaper Seoul Economic Daily, TSMC informed Taiwanese media that the start of volume production of 2 nm has occurred “as scheduled” in the fourth quarter, and that it is the most advanced technology in the sector in terms of transistor density and energy efficiency.
This same report adds a key data point summarizing the current situation: 2 nm production capacity is already approximately booked for a year due to the surge in orders from clients like Apple and NVIDIA, indicating a saturation scenario for anyone arriving later in the queue.
The practical outcome, as always with cutting-edge technology when demand exceeds supply: prices rise and buyers seek diversification. Specifically, Taiwanese media estimate that the cost of 2 nm could exceed $30,000 per wafer, a figure that, while an estimate, illustrates why some clients might try to distribute orders across multiple fabs.
What improvements does jumping to 2 nm bring?
The N2 node from TSMC presents a significant step over 3 nm: the South Korean media reports that, compared to the previous node, the switch promises 25–30% more energy efficiency and 10–15% higher performance, along with the adoption of Gate-All-Around (GAA) architecture to continue scaling once traditional approaches start to hit limits.
From a business perspective, these improvements are golden for two sectors fueling today’s market:
- Premium mobiles, where each efficiency gain translates into longer battery life or sustained performance.
- Artificial Intelligence and HPC, where power consumption (and electricity costs) increasingly impact the total cost of ownership (TCO).
The “Samsung factor”: when the second becomes attractive
The same analysis positions Samsung in a unique place: not leading the race, but benefiting from the leader being overwhelmed and more expensive.
Samsung is advancing its SF2 process and, according to published data, is already producing the Exynos 2,600 (mobile AP) on that node. In their own comparative figures, Samsung notes more modest improvements: compared to their second-generation 3 nm, they mention +8% efficiency and +5% performance.
This isn’t about headline wars but a very specific question for major chip designers (fabless): Is it worth paying more and waiting, or accepting a “good enough” alternative with better availability and negotiation power?
Quick comparison (based on disclosed figures)
| Node | Company | Improvements over previous node (per reports) | Key point |
|---|---|---|---|
| N2 | TSMC | +25–30% efficiency, +10–15% performance | GAA; very high demand |
| SF2 | Samsung | +8% efficiency, +5% performance | Key: demonstrating yields |
Geopolitics and regulation: the “N-2” rule and U.S. manufacturing pressure
Add a political twist to the industry equation. The Seoul Economic Daily reports that the Taiwanese government may have introduced a rule known as “N-2” to limit the export of critical technologies, which could hinder TSMC’s ability to immediately ramp up its most advanced processes at its Arizona plant. In this context, Samsung’s upcoming factory in Taylor (Texas) could become more attractive for U.S. companies seeking leading-edge manufacturing on American soil.
This is significant because it shifts the conversation from “who makes better” to where it’s made and under what restrictions.
Potential clients list: Qualcomm, Meta, Google, AMD… and the domino effect
The South Korean article notes that, with TSMC raising prices and capacity fully booked, Qualcomm would be one of Samsung’s top targets, especially since the chip designer has historically employed a strategy of multiple foundries. It’s also suggested that part of the production for its next flagship mobile chip (referred to as Snapdragon 8 next gen) might be split if costs per wafer keep increasing.
Potential additions include Meta (for its MTIA accelerators) and Google (for its Tensor family, with a history of manufacturing in Samsung), plus AMD, whose interest in dual-sourcing strategies could benefit Samsung if they demonstrate stable 2 nm production.
And, just to underline that this window of opportunity is real, the report highlights that Samsung has already secured significant deals, including orders related to Tesla (autonomous driving chips), Ambarella (ADAS chips), and mentions Preferred Networks (PFN) as an example of AI accelerator traction.
Meanwhile, tech outlets like Wccftech echo this “spill-over” thesis: when TSMC encounters capacity constraints, the industry is pushed to explore alternatives, and Samsung emerges as a prime candidate.
What lies ahead in 2026: less technological romance, more industrial pragmatism
The takeaway from early 2026 is less about epic breakthroughs than about market realities:
- TSMC sets the technological pace, but its success creates a capacity problem: there isn’t enough for everyone, at least short term.
- Samsung can turn that pressure into business if it proves robust yields and timely delivery.
- For clients, the goal isn’t just “the best node” anymore but guaranteed supply, cost control, and reducing dependency risk.
In short: moving to 2 nm doesn’t just mark a new miniaturization phase; it also heralds a new era where competitive advantage may hinge as much on engineering as on availability.
Frequently Asked Questions
Why is there so much talk about “2 nanometers” if it doesn’t always correspond to a real physical size?
Because “node” is mainly an industry label that summarizes a set of improvements (density, power, performance, design rules). It’s not necessarily the literal size of a transistor.
What does it mean when production is “reserved” for a year?
A significant part of future manufacturing capacity is already committed to customer orders. For other buyers, this usually translates to longer lead times or higher prices.
Why can wafer prices skyrocket at cutting-edge nodes?
Because complexity increases, the number of process steps grows, equipment demands intensify, and competition for capacity in the most advanced fabs escalates.
What would Samsung need to demonstrate to secure more 2 nm orders?
Primarily, stable manufacturing yield, sustained volume, and reliable delivery timelines. Achieving this could make “Plan B” strategies a standard consideration for many chip designers.