NVIDIA Backs Transceiver Optics While GPU Optics Remains Open

These commitments highlight optical transceivers as a constraint. As the number of GPUs increases in AI data centers, the amount of optical transceivers required to move data between servers and switches increases. By investing directly in incumbent photonics manufacturers, NVIDIA is reinforcing the existing optical supply chain that supports this expansion. For companies introducing emerging optical architectures, this reinforcement carries competitive implications.

How optical data movement works today

The optical supply chain NVIDIA is reinforcing sits within the transceivers that connect servers and switches. Lumentum and Coherent produce these optical transceivers, converting electrical signals into optical signals for data movement between servers and switches, upstream of the GPU.

Competitive Pressure on Emerging Optical Architectures

While optical transceivers efficiently move data between servers and switches, GPUs increasingly hit a copper bandwidth bottleneck as incumbents operate exclusively at the transceiver level. POET attempts to address this with its optical interposer, enabling optical data movement at the GPU and bypassing the copper bottleneck.

At the same time, NVIDIA’s deepening ties with incumbent photonics manufacturers introduces competitive pressure. Incumbent manufacturers now benefit from capital backing, engineering collaboration, and supply commitments from the largest AI hardware buyer. Over time, they could leverage these advantages to move their technologies closer to the GPU, encroaching on POET’s domain.

Why Incumbent Strength Does Not End the Opportunity

NVIDIA’s investment in incumbent photonics manufacturers does not diminish emerging optical architectures though. The optical hardware produced by Lumentum and Coherent is primarily deployed in optical transceivers connecting servers and switches, enabling high-speed communication between them rather than interfacing directly with the GPU.

The architectural challenge POET addresses sits deeper in the data movement. Its optical interposer targets optical data movement at the GPU, where conventional copper interconnects hit scaling limits. As long as incumbent optical technologies remain concentrated in transceivers between servers and switches, their strength does not extend to the GPU, where POET operates.

Whether Incumbents Extend Into POET’s Domain

NVIDIA’s decision to fund Lumentum and Coherent strengthens the suppliers that already dominate optical data movement between servers and switches in AI data centers. This deepens alignment with incumbent manufacturers that have established manufacturing scale, proven supply chains, and existing integration within NVIDIA’s systems.

For POET, the implication is not displacement, but a more competitive environment shaped by incumbents that are now more tightly coupled to NVIDIA’s roadmap. The relevance of POET depends on whether optical data movement moves closer to the GPU and, if so, whether NVIDIA extends its incumbent partnerships into GPU optics or adopts emerging optical architectures.

Disclosure: This article reflects the author’s personal analysis and opinions and is not investment advice. The author does not hold shares in POET Technologies Inc. at the time of writing. Images used are independent illustrative renderings and are not official POET Technologies Inc. promotional materials.

RISK PROFILE
Incumbent Encroachment: NVIDIA’s backing of incumbent photonics manufacturers supplying optical transceivers strengthens the existing optical supply chain, which could eventually encroach on POET’s optical interposer, shaping the competitive environment for POET.

Leave a Comment