Samsung Bets on Silicon Photonics: Can 2027 Become the Turning Point Against TSMC?

SpotPulse

Spot
Samsung has declared silicon photonics its core breakthrough strategy and is accelerating efforts to commercialize CPO technologies by 2027, leveraging TSMC’s rising prices and capacity burdens to strengthen its competitive position.

Pulse
As AI system bottlenecks shift toward interconnect speed, power efficiency, and thermal constraints, the foundry race is expanding beyond node scaling.

Global customers will increasingly favor vendors that can deliver photonics-enabled packaging, predictable cost structures, and diversified supply resilience—making 2027 a critical inflection point for both Samsung and the industry.

Samsung Electronics has publicly identified silicon photonics as the potential “game changer” that could redefine its long-term competitiveness in the AI semiconductor race.

Korean media note that Samsung is now launching a full-scale effort to narrow its gap with TSMC by accelerating development of co-packaged optics (CPO) and positioning photonics as the foundational technology that can simultaneously address speed, heat, and power bottlenecks in AI chips.

Silicon photonics replaces electrical signal transmission with light. Under today’s copper-wiring architecture, resistance slows data movement and generates additional heat and power consumption.

By contrast, light suffers no resistance, allowing far faster transmission, lower temperatures, and significantly reduced power requirements. For AI chips—where the core challenges now lie in both performance scaling and energy efficiency—this technology is emerging as an essential breakthrough.

Global leaders such as NVIDIA, Intel, and AMD are already moving into photonics-enabled architectures. TSMC, which fabricates many of the world’s leading AI chips, currently holds the technological advantage thanks to its integration experience with major customers.

However, Samsung is now moving aggressively, viewing silicon photonics as a decisive lever in the next phase of foundry competition.

Samsung’s push is anchored in Singapore, where the company is recruiting specialists from TSMC and Intel to accelerate development.

Samsung is targeting 2027 as the first meaningful commercialization window for silicon photonics, indicating that photonics-enabled CPU architectures may emerge as early as that period.

Parallel to this, Samsung is showing notable progress in its 2 nm process. Although the initial ramp suffered yield challenges similar to those experienced at 3 nm, current yield levels for 2 nm are reported to have reached approximately 55–60 percent—close to mass-production readiness.

Production capacity is expanding aggressively: Samsung is expected to move from 8,000 wafers per month in 2024 to 21,000 wafers per month by the end of 2025, a 163 percent increase.

Customer traction is emerging as well. Reports indicate that companies ranging from Tesla to Apple, as well as mobility-focused AI and autonomous driving startups, are selecting Samsung’s advanced nodes.

Meanwhile, TSMC is facing customer discontent as it increases 2 nm wafer prices by more than 50 percent, raising the possibility of partial customer migration.

The combination of high prices, tight capacity, and delivery delays at TSMC is creating openings that Samsung is already attempting to exploit with more flexible pricing strategies.

Below is a structured comparison that incorporates the competitive themes between Samsung and TSMC.

Korean industry experts increasingly focus on 2027 as Samsung’s potential turning point. Tesla’s AI compute volumes are expected to be more fully reflected in Samsung’s foundry business around this time, and the company’s silicon photonics roadmap aligns with this window.

From a financial standpoint, Samsung Foundry is expected to reduce its current KRW 6–7 trillion loss in 2024 to approximately KRW 1–2 trillion in 2025, with a return to profitability potentially beginning in 2027. In addition, the ongoing capacity strain at TSMC—combined with both delivery risk and rising price pressure—has led analysts to conclude that Samsung is increasingly well positioned to draw in customers looking for more stable long-term alternatives.

Ultimately, the competitive landscape is shifting. Silicon photonics and 2 nm yield stabilization will determine whether Samsung can transform these emerging opportunities into sustained share gains.

The main strategic question is whether Samsung can achieve stable commercialization of photonics and full 2 nm yield optimization in time to leverage the market dislocation caused by TSMC’s capacity overload. This will define Samsung’s medium- to long-term trajectory in the AI foundry market.

CXL Is the Next HBM: How Samsung Is Opening the Era of Shared Memory

Want deeper insights?

This SpotPulse® provides only a snapshot of the issue.   Access the full CoreBrief® and InDepth report® for in-depth analysis, data charts, and strategic implications tailored for decision-makers. Contact@AIStrategica.com