The "A20 Pro" destined for the iPhone 18 Pro is predicted to combine TSMC's 2nm-class "N2" process with wafer-level multi-chip module (WMCM) packaging. The 2nm adoption was first predicted in September 2024, and the move to WMCM was reported in August 2025. Apple has not yet announced either the product or the chip. Even so, it's notable that predictions about switching both the manufacturing process and the packaging process in the same generation have converged. The former governs transistor performance and power, while the latter determines how multiple components are assembled together. Looking at these two elements separately makes clear what we can expect from the A20 Pro—and what we'll only know once we see the actual hardware.
What 2nm Changes About Transistor Options
TSMC's N2 entered mass production in Q4 2025. It's the company's first process to use nanosheet transistors, meaning the transistor structure itself differs from the previous 3nm generation. If the A20 Pro adopts N2, it would mark the first 2nm-generation chip for iPhone. Analyst Ming-Chi Kuo and others have offered this prediction.
According to TSMC's own comparison with N3E, N2 delivers 10-15% higher speed at the same power, or reduces power consumption by 25-30% at the same speed. Chip density is also said to increase by more than 15%. Designers can choose how to allocate this headroom—toward peak performance, battery life, or circuit scale.
However, these figures come from TSMC's standard-condition process comparisons, not from a performance comparison between the A20 Pro and A19 Pro. The actual gains realized in the product depend on how Apple configures the CPU and GPU, sets clock speeds, and determines die size. Even if N2 adoption is confirmed, it doesn't automatically produce an iPhone that's "15% faster" or "30% more power-efficient."
From InFO to WMCM: Back-End Packaging Also Shifts
According to Kuo's supply chain research, the A20 for the iPhone 18 lineup in the second half of 2026 will switch packaging methods from the conventional InFO to WMCM. If the process is the technology that builds transistors on silicon, packaging is the technology that connects and protects the finished die, turning it into a component that can be mounted in a device. If both N2 and WMCM are adopted, major changes will occur in both front-end and back-end processes.
One specific change the report cites is the adoption of MUF (Molding Underfill). MUF combines two steps into one: the underfill process that fills gaps between dies and protects connection points, and the molding process that encapsulates everything in resin. Kuo states that this can reduce material usage and process steps, improving yield and production efficiency.
The advantage here shows up in manufacturing before it shows up in computational performance. If fewer process steps mean more consistent yields of good chips, it becomes easier to scale the new packaging approach for mass iPhone production. On the other hand, it hasn't been disclosed which dies are separated within the A20 Pro or how DRAM is connected. The name "WMCM" alone doesn't tell us how much AI processing or gaming performance will improve.
Look at Heat and Power Before AI Performance
The current iPhone 17 Pro pairs the A19 Pro with a vapor chamber. Apple states that this combination delivers up to 40% higher sustained performance compared to the previous generation. The chip alone doesn't support that peak figure—a mechanism that dissipates generated heat into the chassis and maintains clock speeds over extended periods is what makes it possible.
If N2's power efficiency carries over to the A20 Pro, it could enable designs that accomplish the same tasks with less power while suppressing heat generation. As for WMCM, the benefits confirmed in the original report are limited to reducing materials and process steps and improving yield and production efficiency. Since the die layout and connection methods haven't been disclosed, the effects on heat and power remain undetermined.
For on-device AI, what shapes usability isn't instantaneous computational speed but power consumption and heat during sustained processing. The A19 Pro equips each core of its 6-core GPU with a Neural Accelerator and also includes a 16-core Neural Engine. Evaluating the A20 Pro will require more than just comparing core counts—it will need measurements, under identical conditions, of speed degradation, battery drain, and surface temperature during continuous execution of local models.
The Final Unknowns: Adoption Scope and Real-World Measurements
In September 2024, Kuo suggested that cost considerations might limit 2nm chip adoption to the iPhone 18 Pro lineup. Subsequent research reports that the A20's packaging will shift to WMCM in the second half of 2026. Whether N2 and WMCM will be adopted uniformly across the same device lineup, or whether specifications will differ between the standard model and the Pro, won't be confirmed until Apple makes an official announcement.
While Apple disclosed the A19 Pro's CPU, GPU, and Neural Engine configuration, it did not list the manufacturing node in its spec sheet. There's no guarantee that the A20 Pro's official announcement will confirm names like N2 or WMCM either. Determining the adopted technologies with certainty will require chip analysis and teardown investigations in addition to product specifications.
The numbers to watch for the 2026 new iPhone aren't TSMC's general process metrics—they're the sustained performance and power consumption of devices equipped with the A20 Pro. N2's actual effect will be confirmed through real-device measurements, while WMCM's implementation will be verified through teardowns and chip analysis.