Classical Optics and the AI Photonics Revolution

Classical optics is the study of light and its behavior, developed over centuries by scientists from Newton to Maxwell. It covers reflection, refraction, diffraction, and interference, the fundamental principles that describe how light travels and interacts with matter. For most of its history, classical optics was concerned with telescopes, microscopes, and cameras. Today, those same principles are being applied to solve one of the most pressing engineering challenges in the world: moving data inside AI data centers.

Jason Bodner’s Accelerated AI presentation is fundamentally about applying classical optics to AI infrastructure. He calls it photonics, and the concept is simple: use light instead of electricity to transmit data between processors. The physics behind this is classical optics, refined and miniaturized onto silicon chips.

How Classical Optics Applies to AI Data Centers

Inside an AI data center, thousands of GPUs need to communicate. The data currently moves over copper wires. The problem is that electrons in copper are slow, generate heat, and consume significant power. Light, governed by the principles of classical optics, solves all three problems.

Reflection and total internal reflection. Fiber optic cables use total internal reflection to channel light through a glass or plastic core with minimal loss. The light bounces along the fiber, staying confined, even around curves. This is the same principle that Newton described, applied at microscopic scale inside AI data centers.

Refraction and modulation. Optical modulators use the principle of refraction to encode data onto light. By varying the refractive index of a material, they can change the intensity or phase of a light beam, effectively turning it on and off at billions of times per second. This is how data is encoded onto light inside a photonic chip.

Wavelength division multiplexing. One of the most powerful applications of classical optics in photonics is wavelength division multiplexing (WDM). Because light of different wavelengths (colors) does not interfere with each other, multiple data streams can be transmitted over the same fiber simultaneously. This multiplies the effective bandwidth of a single optical connection, something that is physically impossible with copper.

Broadcom’s 400G Optical Chip

Broadcom debuted the industry’s first 400G optical chip, which applies these classical optics principles at scale. The chip converts electrical signals from GPUs into light pulses, transmits them over fiber, and converts them back to electrical signals at the destination. The 400G designation means it handles 400 gigabits per second, enough to transfer a high-definition movie in a fraction of a second.

Broadcom’s AI revenue was $10.8 billion last quarter, up 143 percent year over year. The company also designs custom AI silicon for OpenAI, Meta, Anthropic, and Google, and builds the networking switches that let AI chips communicate. For more on Broadcom, see our Broadcom stock article. For more on the optical chip, see our Broadcom optical chips article.

The Acceleration Curve: Optics in Historical Context

Bodner frames the application of classical optics to AI inside his Acceleration Curve framework. The pattern has played out twice before:

  • Internet (1990s): Copper phone lines replaced by fiber optics. Classical optics principles applied to telecommunications. Equinix rose 37,000 percent, Netflix rose 188,471 percent.
  • Smartphones (2007): Copper to fiber in mobile networks. Monolithic Power rose 17,000 percent.
  • AI (now): Copper to photonics inside data centers. Bodner says the second wave will be bigger.

The pattern is consistent. A technology goes through an initial boom, hits a copper bottleneck, and then classical optics provides the breakthrough that enables the next phase of growth. For more on the framework, see our copper to fiber article.

The Industry Investment

The application of classical optics to AI is not speculative. The largest companies in the world are investing billions:

  • Nvidia has invested over $7 billion in photonics companies and partnered with Corning to build three optical factories.
  • AMD is building a $280 million photonics research hub.
  • Ayar Labs raised $500 million from ARK Invest and Sequoia Capital.
  • Bill Gates has personally invested over $200 million in two photonics companies.
  • The Optical Interconnect Alliance includes Nvidia, AMD, Broadcom, Microsoft, Meta, and OpenAI.

Sequoia Capital called photonics “holy grail tech.” For more on the alliance, see our optical interconnect article.

Considerations

Classical optics is a well-established science. The principles are not in question. What is new is the application of those principles at the scale and precision required for AI data centers. The engineering challenges are significant, but the investment flowing into solving them suggests the industry is committed to making it happen.

Bodner’s track record, including Nvidia at $4.50 and Super Micro before ChatGPT, suggests he has a genuine ability to identify when a technology shift is about to accelerate. For more on his background, see our Jason Bodner profile. For the full thesis, see our Accelerated AI review.

If you want to explore the full thesis, you can access the Accelerated AI presentation through Brownstone Research.

This is not financial advice. Always do your own research before investing.