Ayar Labs Raises $500M to Power AI With Light
Ayar Labs closes a $500M Series E backed by NVIDIA, AMD & Sequoia to replace copper with optical tech in AI data centers. Read the full AI funding news here.
TL;DR
Ayar Labs, a San Jose-based semiconductor startup, has raised $500 million in Series E funding at a $3.75 billion valuation, bringing its total funding to $870 million. The round was led by Neuberger Berman and backed by NVIDIA, AMD, Sequoia Capital, ARK Invest, MediaTek, and the Qatar Investment Authority. The company replaces traditional copper wiring inside data centres with optical connections that move data using light, delivering up to 10x lower latency and 5–10x higher bandwidth than copper — directly addressing the biggest bottleneck slowing down large-scale AI infrastructure today. Funds will be used to scale manufacturing and open a new office in Taiwan.
Ayar Labs Closes $500M Series E Funding Backed by NVIDIA and AMD to Replace Copper with Light in AI Infrastructure
The artificial intelligence hardware race just witnessed one of its most significant milestones in early 2026. Ayar Labs, a Silicon Valley-based semiconductor startup pioneering optical interconnect technology, has successfully closed a massive $500 million Series E funding round, pushing its valuation to an impressive $3.75 billion. This landmark AI funding news comes at a time when the entire technology industry is grappling with one of the most stubborn challenges in modern computing — moving data fast enough to keep up with the explosive growth of AI systems. The round was led by Neuberger Berman, with co-leadership from Insight Partners, and included strategic participation from some of the biggest names in the semiconductor world, including NVIDIA, AMD, MediaTek, the Qatar Investment Authority, Alchip Technologies, ARK Invest, Sequoia Capital, and 1789 Capital. With this new injection of capital, Ayar Labs' total funding now stands at an impressive $870 million — cementing the company's position as the undisputed leader in co-packaged optics (CPO) solutions designed specifically for large-scale AI deployments.
For those who follow AI funding news closely, this development represents far more than just another large cheque being written in Silicon Valley. It reflects a growing industry-wide consensus that the age of copper interconnects inside data centres is drawing to a close — and that the future of AI infrastructure will be built on light, not electrons. Ayar Labs sits right at the heart of that transition, having spent more than 15 years developing the core technologies needed to make this shift possible. The company's timing could not be better, as the global AI infrastructure buildout enters a phase where raw computational power is no longer the only constraint. How data moves between chips — and how much power that movement consumes — has become just as important as the chips themselves.
The Problem That Copper Can No Longer Solve
To appreciate why Ayar Labs' $500 million raise is such a big deal, it helps to understand the fundamental problem the company is trying to solve. For decades, copper wiring has been the standard way to connect semiconductors inside server racks. Copper is cheap, well-understood, and deeply embedded in the supply chains and manufacturing processes that power the global data centre industry. However, as AI systems have grown to require increasingly massive clusters of processors working in tight coordination, copper's physical limitations have become impossible to ignore.
The core issue is simple physics. When electrical signals travel through copper wires at high speeds, signal quality degrades. The faster you push data through copper, the more energy is lost as heat, and the more the signal quality deteriorates. This means engineers are forced to make uncomfortable trade-offs between speed, power consumption, and data integrity. In the context of modern AI training and inference workloads — which require thousands of graphics processing units and AI accelerators to exchange enormous volumes of data in real time — these trade-offs add up to significant performance bottlenecks that limit both the scale and the efficiency of AI systems.
Gabe Cahill, Managing Director at Neuberger Berman, captured the situation plainly when he noted that companies tend to stay with copper for as long as they possibly can, but that pushing more electrons through copper inevitably reduces signal fidelity. He described the transition to using light for data transmission as a genuinely significant step forward — and one that Neuberger Berman is committed to backing all the way through to what the firm sees as a likely near-term initial public offering. Cahill also noted that the current expansion of AI infrastructure represents one of the most significant capital investment opportunities of our time, with data centre interconnect emerging as the primary bottleneck holding the industry back. It is precisely in this context that the AI funding secured by Ayar Labs takes on its full strategic significance.
How Ayar Labs Is Solving AI's Biggest Hardware Bottleneck
Ayar Labs operates in a highly specialised area of semiconductor technology known as co-packaged optics, which is itself a subset of the broader field of silicon photonics. The company's central innovation is elegantly straightforward in concept, even if fiendishly difficult to execute in practice: instead of moving data as electrical signals through copper wires, Ayar Labs' chips send and receive data as pulses of light through optical fibre connections. This approach, known as optical interconnect technology, delivers transformative improvements in both speed and energy efficiency.
At the heart of Ayar Labs' technology is the TeraPHY optical I/O chiplet — a compact, highly integrated component that can deliver up to 8 terabits per second of bandwidth while conforming to the UCIe (Universal Chiplet Interconnect Express) open standard. This ensures that TeraPHY can be seamlessly integrated into AI accelerators and network switches built by a wide range of manufacturers. The TeraPHY platform is paired with the SuperNova external light source, a 16-wavelength solution that enables unmatched bandwidth density across large-scale AI systems. Together, these technologies give Ayar Labs' customers access to optical interconnects that deliver between four and twenty times more computing throughput per watt compared to traditional copper-based solutions, along with five to ten times higher bandwidth, four to eight times better power efficiency, and ten times lower latency.
Mark Wade, the co-founder and chief executive officer of Ayar Labs, has consistently described the company's mission in direct terms. "We're solving one of the biggest hardware issues that's causing bottlenecks in AI," he said, adding that the company anticipated many years ago that copper connectivity would eventually limit computing performance during the 2020s and 2030s. That prediction has now come true, and Ayar Labs finds itself in the enviable position of having a mature, production-ready solution at exactly the moment the market needs it most. The company has announced plans to use the Series E proceeds to scale its high-volume manufacturing capabilities, expand its test capacity, and establish a new office in Taiwan — where much of the world's most advanced semiconductor packaging work takes place in partnership with TSMC.
The Heavyweights Behind This AI Funding Round and What It Signals
The quality and composition of the investor syndicate in Ayar Labs' latest AI funding round is arguably as newsworthy as the size of the raise itself. The participation of NVIDIA and AMD — the two dominant forces in AI accelerator hardware — is particularly telling. Both companies have enormous vested interests in ensuring that the data centre infrastructure supporting their chips operates as efficiently as possible. NVIDIA, for instance, has long relied on its proprietary NVLink high-speed copper connections to link chips within data centres, and has invested continuously in improving those systems to support ever-more-powerful processors. However, as AI clusters have grown larger and inference demand has skyrocketed, even NVIDIA's best copper solutions are beginning to show their limits. By backing Ayar Labs, NVIDIA is effectively hedging its bets and ensuring it has a stake in the optical interconnect technology that may ultimately replace or augment its own copper-based systems.
The involvement of the Qatar Investment Authority alongside technology-focused investors like Sequoia Capital and ARK Invest underscores the global and cross-sector appeal of Ayar Labs' technology. Sovereign wealth funds backing deep-tech semiconductor startups is a relatively recent phenomenon, but it reflects the growing recognition that AI infrastructure hardware is no longer a niche concern — it is a matter of national economic competitiveness and strategic importance. Sequoia Capital's participation brings the weight of one of Silicon Valley's most storied venture capital firms, which has a long history of identifying and backing the companies that define technological transitions.
Perhaps the most intriguing figure associated with Ayar Labs is Pat Gelsinger, the former chief executive officer of Intel, who now serves as a board member at the company. Gelsinger has noted that he began researching silicon photonics more than two decades ago, and that early predictions about the rapid demise of copper interconnects turned out to be premature. However, he now believes that the extraordinary scale of modern AI infrastructure — with hyperscaler data centres housing hundreds of thousands of processors connected in massive clusters — has fundamentally changed the equation. Gelsinger said Ayar Labs has addressed the technical manufacturing challenges that stalled the industry for years, and that the company is now genuinely prepared for high-volume production. His endorsement carries significant credibility given his decades of experience at the frontier of semiconductor technology.
The Broader AI Photonics Race and Ayar Labs' Competitive Edge
Ayar Labs' $500 million Series E is not happening in isolation. It is part of a broader wave of AI funding flowing into the silicon photonics and optical interconnect space, as the industry collectively comes to terms with copper's limitations. This AI funding news arrives just weeks after NVIDIA made multi-billion-dollar commitments to optical component manufacturers Lumentum and Coherent, signalling that the world's most valuable AI company is actively preparing for a post-copper future in its data centre deployments. In February 2026, semiconductor giant Marvell completed a $3.25 billion acquisition of Celestial AI, a photonic interconnect startup, further confirming that optical solutions for AI infrastructure are no longer a future possibility — they are an imminent commercial reality.
Within this competitive landscape, Ayar Labs has carved out a distinctive position. Unlike some competitors who are developing standalone optical networking solutions, Ayar Labs focuses specifically on co-packaged optics — integrating optical components directly alongside silicon chips in the same package. This approach eliminates the long copper traces that connect chips to optical transceivers in traditional designs, dramatically reducing both the latency and the power consumption associated with optical data transmission. The company's TeraPHY chiplet is already being validated by customers and strategic partners, including Alchip Technologies, with which Ayar Labs has a deep integration partnership using TSMC's advanced packaging technologies, including COUPE and TSMC-SoIC processes.
The company is also working with Intel Foundry and has demonstrated interoperability with the UCIe standard, which is becoming the industry's preferred framework for connecting multiple chiplets from different manufacturers within a single package. This positions Ayar Labs not just as a component supplier, but as a foundational technology provider within the broader chiplet ecosystem that is increasingly underpinning the design of next-generation AI accelerators. For hyperscaler customers planning their 2027 and beyond data centre deployments, Ayar Labs' roadmap aligns directly with their most pressing performance and efficiency requirements.
From AI Training to Inference: Why This Transition Demands Optical Solutions Now
One of the most important contextual factors driving demand for Ayar Labs' technology is the shift in the AI industry's centre of gravity from training to inference. For most of the past decade, the dominant computational challenge in AI was training large language models and other foundation models — a process that involves processing enormous datasets over extended periods to build the statistical patterns that give AI systems their capabilities. Training is computationally intensive, but it is a relatively batch-oriented process where latency matters less than raw throughput.
Inference is a different beast entirely. Inference is what happens when a trained AI model is deployed in the real world to respond to user queries — powering chatbots, AI agents, content generation tools, and a vast array of applications that hundreds of millions of people now use on a daily basis. Inference demands fast, consistent, low-latency responses, and it happens at enormous scale. Every query to an AI assistant, every AI-generated image, every autonomous agent making a decision — all of these require chips to exchange data rapidly and repeatedly. This is precisely where the physical limits of copper wiring become most painfully apparent, and precisely where optical interconnects offer the most compelling advantages.
As AI companies race to scale their inference infrastructure to meet soaring demand, the efficiency of data movement between chips inside their data centres has become a direct determinant of how many queries they can handle per second and at what cost. Every watt saved on data movement is a watt that can be redirected to computation. Every reduction in latency translates to faster, more responsive AI services. The economic and operational incentives for transitioning from copper to optical interconnects are, in this context, overwhelming — and they only grow stronger as AI clusters continue to scale in size and complexity. This is why AI funding of the magnitude secured by Ayar Labs is not just justified but arguably overdue, and why the company's investors clearly believe they are backing a technology that will be at the heart of AI infrastructure for years to come.
Ayar Labs' journey from a research concept to a company on the verge of high-volume manufacturing is a testament to what sustained, patient capital and deep technical expertise can achieve. With $870 million in total AI funding raised, a valuation of $3.75 billion, and the backing of the semiconductor industry's most powerful players, the company is entering its most consequential phase yet — one where laboratory achievements must translate into mass-market products deployed in the world's most demanding data centre environments. The road ahead is challenging, but the industry's direction is clear: AI's future runs on light, and Ayar Labs intends to be the company that delivers it.