The AI optical interconnect sector welcomes an epic integration! A $115 billion super merger endorsed by Nvidia (NVDA.US) CEO Jensen Huang

English Financial APP reports that chip maker Marvell Technology (MRVL.US) announced on Monday that it will acquire data center laser manufacturer Lumentum Holdings (LITE.US) in an all-stock deal valued at approximately $115 billion. If successful, this transaction, based on equity value, will not only be the world’s largest merger since 2026, but also the largest in the global tech industry aside from the merger of SpaceX (SPCX.US) and xAI.

Prior to finalizing this massive deal, the leaders of both Marvell Technology and Lumentum had received support from Nvidia (NVDA.US) CEO Jensen Huang. According to people familiar with the matter, Marvell CEO Matthew Murphy and Lumentum CEO Michael Hurlston sought Huang’s opinion over the phone before the transaction announcement.

Huang’s endorsement highlights that as demand for faster optical connections becomes a bottleneck in increasingly large data center construction, Nvidia, the leader in artificial intelligence (AI), is driving integration among key suppliers. Michael Hurlston stated in an interview Monday, “I think Nvidia’s response has been very positive. We believe that through this merger, we can actually deliver more products to Nvidia faster.”

Matthew Murphy said during an investor conference call discussing the deal, "Mike and I share a very similar vision for the direction of the world." "We have maintained a longstanding relationship both professionally and personally—Lumentum is a major customer of Marvell Technology. We interact with the Lumentum team a lot. All of this came together naturally."

What are the benefits of the Marvell Technology and Lumentum merger for Nvidia?

Earlier this year, Nvidia invested $2 billion each in Marvell Technology and Lumentum, totaling $4 billion. The merger of these two companies may not only bring investment returns to Nvidia, but also has the following important implications for Nvidia.

1. Unblocking the high-speed optical interconnect supply chain, easing the AI data center connectivity bottleneck

The larger the AI computing scale, the more likely data transmission becomes a bottleneck. Nvidia needs not only more powerful GPUs, but also high-speed interconnect systems that enable efficient collaboration among tens of thousands of GPUs. This deal happens to integrate the companies’ technologies in electrical interconnects, optical interconnects, and custom chips.

The combined company can merge Marvell's chip, signal processing, and interconnect technologies with Lumentum's lasers, optical devices, and manufacturing capacity. This means it could transition from supplying a single chip or optical component to providing more complete high-speed interconnect solutions. For Nvidia, this helps improve the supply of optical devices for AI data centers and drives overall performance enhancement for next-generation AI infrastructure.

2. Accelerating the adoption of silicon photonics and next-generation optical interconnect technologies

Silicon photonics is another focus of this transaction. Simply put, silicon photonics combines optical components with semiconductor chip manufacturing processes, enabling data transmission between chips and systems via optical signals. It is expected to solve the power consumption and bandwidth limitations faced by traditional electrical interconnects in high-speed transmission scenarios.

Nvidia has previously invested in this area with both Marvell Technology and Lumentum. In cooperation with Marvell, both parties are advancing silicon photonics technology and connecting Marvell’s custom XPU and network technologies to the Nvidia ecosystem via NVLink Fusion. In partnership with Lumentum, both sides jointly develop advanced optical technologies, focused on lasers and optical connectivity solutions needed for next-generation AI infrastructure.

After the merger, the two companies are expected to further connect optical devices, semiconductor chips, signal processing, and packaging, shortening product development cycles. For Nvidia, this means that as it builds ultra-large-scale AI clusters in the future, it may obtain more mature, highly integrated optical interconnect solutions.

3. Strengthening supply chain security and reducing risks in the supply of critical optical components

As the scale of AI infrastructure construction expands, supply chain bottlenecks are no longer limited to GPUs themselves but also include high-bandwidth memory, advanced packaging, network switches, optical transceivers, and lasers. High-performance lasers, in particular, are a vital part of optical communication systems and directly affect optical signal generation and transmission capabilities.

Nvidia previously invested $2 billion in Lumentum, securing long-term procurement commitments and future capacity usage rights. This means Nvidia wishes not only to participate in technology R&D but also hopes to lock in supply capacity for key products ahead of time.

If Marvell Technology and Lumentum successfully merge, the new company may further coordinate chip design, optical device production, and system-level product development. This will help Nvidia reduce supply chain coordination costs and improve the delivery certainty of next-generation AI data center products.

4. Promoting the expansion of Nvidia’s AI ecosystem and enhancing platform competitiveness

This deal may also have a less recognized significance: Nvidia is driving its shift from a GPU supplier to a comprehensive AI infrastructure platform. Previously, Nvidia has developed a complete technology system covering computation, networking, and data transmission through products such as NVLink, NVLink Fusion, Spectrum-X, ConnectX, and BlueField. Among them, NVLink Fusion allows customers to develop customized AI computing infrastructures while maintaining compatibility with Nvidia’s technology ecosystem.

Marvell Technology is a key partner in this strategy. The collaboration allows Marvell to provide custom XPU and high-speed interconnect technology compatible with NVLink Fusion, while Nvidia offers Vera CPU, ConnectX network cards, BlueField DPU, NVLink interconnect, and Spectrum-X switches.

Lumentum’s addition is expected to further enhance the ecosystem’s optical connectivity capabilities. From an industry competition perspective, this helps Nvidia respond to rivals such as Broadcom (AVGO.US) in customized AI chips, network switches, and data center infrastructure. Especially as major cloud service providers increasingly focus on customized AI chips, Nvidia needs to ensure that even if clients use non-Nvidia GPUs or custom XPUs, their infrastructure can still utilize Nvidia’s interconnect technology and network ecosystem. Therefore, Nvidia’s support of the Marvell Technology and Lumentum merger aligns with its strategic direction of expanding the coverage of the AI infrastructure ecosystem.