Introduction
The world knows Largan Precision as the supplier of high-end smartphone camera lenses, particularly for Apple. But a recent report from Economic Daily News, citing sources familiar with the matter, reveals that Largan is quietly moving into the AI infrastructure space. The company is reportedly collaborating with TSMC on co-packaged optics (CPO), with samples already under testing in a project involving AMD.
If confirmed, this marks a pivotal moment—not just for Largan—but for the evolution of data center architectures and the global AI supply chain.
techovedas.com/cpo-surge-how-tsmc-mediatek-are-powering-the-future-of-5g-and-ai-chips
5 Quick Takeaways
- Largan goes AI – From iPhone lenses to TSMC’s co-packaged optics.
- AI’s real bottleneck – Data movement, not compute.
- CPO advantage – Optical engines next to ASICs = faster, denser, efficient.
- Industry backing – NVIDIA deployed; AMD tests underway.
- Strategic edge – Precision optics now a core AI infrastructure skill.
Follow us on LinkedIn for everything around Semiconductors & AI
Why AI Data Centers Are Shifting to Optics
For years, AI performance headlines have focused on GPUs, nodes, and transistor counts. But the real bottleneck in modern AI systems isn’t compute—it’s data movement.
As AI clusters scale:
- East–west traffic between accelerators surges
- Copper-based interconnects hit power and thermal limits
- Signal integrity and latency challenges grow exponentially
Traditional copper architectures struggle to support 800G, 1.6T, and next-generation AI fabrics. This is where optical interconnects—and particularly co-packaged optics—become crucial.
CPO moves the optical engine next to the switch ASIC, shrinking electrical traces and handing data to light immediately. The result:
- Lower power consumption
- Higher bandwidth density
- Better scalability for AI clusters
In short, CPO is not an incremental upgrade—it’s a fundamental redesign of data center networking.
Largan’s Expertise Fits Perfectly
CPO needs extreme precision: aligning photonic chips with fibers, redirecting light via prism-based collimators, and maintaining sub-micron tolerances at scale.
Largan’s core strengths match this perfectly:
- Ultra-precision machining
- Automated alignment systems
- High-volume quality control
Through its PhotoniCore Technologies subsidiary, Largan is setting up dedicated CPO production lines, signaling a long-term commitment to AI optics.
TSMC’s Strategic Role in Silicon Photonics
Largan’s involvement underscores TSMC’s evolution beyond chip fabrication.
TSMC’s COUPE (Compact Universal Photonic Engine) platform, verified in early 2025, enables scalable co-packaged optics deployment. It integrates:
- Advanced packaging
- Silicon photonics
- Logic and memory interfaces
By co-packaging optics with ASICs, TSMC is setting the stage for next-generation AI switches, offering power, speed, and density advantages that pluggable modules cannot match.
techovedas.com/what-is-tsmc-foundry-2-0-why-tsmc-is-changing-its-path
NVIDIA and AMD Validation
CPO is already proving itself in the field. NVIDIA’s Spectrum-X switch, adopted by hyperscalers including Oracle and Meta, demonstrates that co-packaged optics can deliver:
- 3.5× better power efficiency
- 10× higher network resiliency
- 1.3× faster deployment than traditional architectures
The report that AMD-linked testing is underway suggests that the industry is moving beyond single-vendor validation.
Co-packaged optics is becoming a cross-platform standard, positioning Largan’s components as a critical enabler.
Investment Signals: Largan Is Serious
Largan is backing its strategic shift with capital. Economic Daily News reports that its 2025 capex is projected to exceed NT$13 billion, up more than 25% year-on-year. Investments are being made in:
- CPO-dedicated production lines
- Core smartphone lens business
- Precision optical automation
This dual-track approach ensures Largan can fund innovation in AI optics while maintaining its traditional revenue streams.
Why Co-Packaged Optics Matters
The shift to CPO reflects a larger reality:
- Data movement now defines AI performance
- Traditional copper interconnects are hitting physical limits
- Hyperscalers require higher density, lower latency, and energy-efficient networking
- Optical manufacturing precision is becoming a strategic asset
- First movers in CPO will control both the supply chain and ecosystem lock-in
Companies that ignore optics risk falling behind, even if they produce leading-edge chips.
Our Take
Largan Precision’s move from iPhone lenses to AI optics is not a simple diversification—it is a strategic pivot into the next bottleneck of computing.
- For India and emerging semiconductor ecosystems: optics and advanced packaging must be priorities. Otherwise, fabs and assembly lines alone will not make the country competitive in AI hardware.
- For investors: suppliers like Largan, with precision manufacturing expertise, may capture outsized margins as hyperscalers scale optical infrastructure.
- For the AI industry: the era of evaluating systems purely on compute or memory speed is over; data movement efficiency is the real performance metric.
Conclusion
Largan Precision move signals a quiet power shift.
In the AI era, the fastest chip doesn’t win—the fastest data movement does.
TSMC, NVIDIA, AMD, and Largan are already shaping the future at the intersection of light, silicon, and precision manufacturing. Miss optics, and you miss the AI infrastructure game.
Subscribe to TechoVedas for expert analysis, industry updates, and in-depth coverage of the chip war and tech geopolitics.
