Forget Oil. These 3 AI Stocks Will Make Millionaires By 2029

Key Concepts

• Optical Networking: The transmission of data using light through fiber-optic cables rather than electrical signals over copper, essential for high-speed AI data center connectivity.
• Bottlenecks in AI: The shift from compute/memory constraints to network bandwidth constraints, exacerbated by the efficiency gains of new data compression methods.
• TurboQuant: A Google DeepMind data compression method that reduces KV cache size by over 80%, accelerating inference by up to 8x.
• Co-packaged Optics (CPO): A design where optical engines are placed directly next to the switch chip, reducing power consumption and signal loss.
• Indium Phosphide (InP): A semiconductor material superior to silicon for generating and detecting laser light in fiber-optic networks.
• Transceivers: Plug-in modules that convert electrical signals to light (and vice versa) at the ends of fiber-optic links.

1. The Shift to Optical Networking

As AI models evolve, the primary constraint in data centers has shifted from raw compute (GPUs) to the speed at which data can be moved.

• The Catalyst: The introduction of TurboQuant allows GPUs to process tokens significantly faster by optimizing memory usage. This creates a "network bottleneck," as the network must now keep pace with the increased throughput of the GPUs.
• Technical Advantage: Optical networking supports speeds of 400G, 800G, and 1.6T (terabits per second), far exceeding the capabilities of traditional copper wiring, which suffers from signal degradation over distance.

2. Market Dynamics and Risks

• Market Growth: The AI optics market is projected to grow from ~$18 billion in 2025 to ~$90 billion by 2030.
• Business Model: These companies rely on heavy upfront R&D and capital expenditure (CapEx) for fabs and manufacturing, leading to multi-year design wins with hyperscalers (e.g., Google, AWS, Meta).
• Shared Risks:
  • Customer Concentration: High reliance on a few major cloud providers.
  • Supply Chain: Vulnerability to export controls, tariffs, and global logistics disruptions.
  • Competition: Significant pressure from tech giants like Broadcom, Nokia, and Huawei.

3. Company Profiles and Strategic Positioning

Lumentum (LIT)

• Focus: Components, specifically high-power lasers and optical switches.
• Key Metric: Revenue grew 65% YoY to $665 million; operating margins expanded from 8.2% to 25.2%.
• Strategic Edge: A $2 billion investment and multi-billion dollar purchase agreement from Nvidia provides capital to build a new US-based fab in North Carolina.
• Application: Leading the transition to co-packaged optics, which reduces power-per-bit by 30–70% by shortening the electrical path.

Coherent (COHR)

• Focus: Vertical integration across the entire photonic stack (from wafers to transceivers).
• Key Metric: Data center segment revenue grew 34% YoY; backlog is 4x higher than current shipment rates.
• Technical Edge: Utilization of 6-inch Indium Phosphide (InP) wafer fabs, which increases chip yield by 4x and reduces die costs by over 60% compared to 3-inch processes.
• Strategic Edge: Like Lumentum, they secured a $2 billion investment and purchase commitment from Nvidia.

Ciena (CIEN)

• Focus: Systems-level integration, stitching components into complete, high-capacity networks.
• Key Metric: $7 billion backlog; cloud provider revenue grew 76% YoY.
• Technical Edge: The WaveLogic 6 solution enables 1.6T per second over a single wavelength, allowing operators to double capacity on existing fiber infrastructure.
• Market Position: Acts as a "systems integrator" that benefits regardless of which specific laser component supplier wins, as they package these parts into end-to-end networks.

4. Synthesis and Conclusion

The transition to AI-driven data centers has made optical networking the critical "plumbing" of the modern internet. While all three companies are well-positioned to benefit from the massive CapEx cycle of hyperscalers, they offer different risk/reward profiles:

• Lumentum is the pure-play component specialist.
• Ciena is the diversified systems-level provider.
• Coherent is identified as the top pick due to its deep vertical integration (owning the InP wafer process) and its ability to capture value at every stage of the optical stack, further de-risked by its strategic partnership with Nvidia.

Final Takeaway: Investors looking for exposure to the AI infrastructure buildout should focus on the "bottleneck solvers." As networking becomes the primary constraint for AI performance, companies that can provide higher bandwidth with lower power consumption—specifically those backed by Nvidia’s capital—are positioned for significant long-term growth.