top of page

Photonic Computing: Why a USD 500 Million Round Points to a USD 8.83 Billion Market

Writer: Strategic Revenue Insights
Strategic Revenue Insights
Nov 10
10 min read
Photonic computing: why a USD 500 million round points to a USD 8.83 billion market, Strategic Revenue Insights

Executive summary. Artificial intelligence clusters have hit a wall that is made of copper. Moving data between thousands of accelerators now consumes a large share of the power and the budget of a data centre, and electrical links are running out of reach and bandwidth. Light is the answer the industry is converging on. Strategic Revenue Insights values the global photonic computing market at USD 550.00 million in 2025 and forecasts USD 8.83 billion by 2035, a 32.0 percent compound annual growth rate and a sixteenfold expansion. The drivers are AI cluster power and bandwidth limits, co-packaged optics adoption, the energy cost of AI inference and foundry photonics platforms. The restraints are analog precision and noise, and packaging and software maturity. This article sets out five lessons from the start-ups, chip majors and hyperscalers moving first, with takeaways for strategy, product, manufacturing, software and revenue.


The structure of the market shows where value is concentrating. Strategic Revenue Insights' Photonic Computing Market research finds that optical interconnects are the largest component at 38.00 percent of 2025 value while photonic processors grow fastest at 34.3 percent, that hybrid electro-optical architectures hold 42.00 percent while spiking photonic neural networks grow fastest at 35.5 percent, that AI training and inference is both the largest and the fastest growing application at 36.00 percent share and 34.5 percent growth, and that North America holds 40.00 percent of the market while Asia Pacific grows fastest at 34.0 percent. The research estimates that Lightmatter leads with 11.5 percent share, followed by Ayar Labs at 8.5 percent, Celestial AI at 7.0 percent, Intel at 6.5 percent, NVIDIA at 6.0 percent, Lightelligence at 5.0 percent, Cisco at 4.0 percent, Q.ANT at 3.5 percent, Salience Labs at 3.0 percent and iPronics at 2.5 percent.


The capital markets have priced the shift. In March 2026 Ayar Labs raised USD 500 million in a Series E round at a USD 3.75 billion valuation, taking its total funding to about USD 870 million, according to Optics and Photonics News. Lightmatter raised USD 400 million in October 2024 at a USD 4.4 billion valuation, according to the company. NVIDIA's Quantum-X Photonics co-packaged optical switches are shipping in 2026, with Spectrum-X Photonics Ethernet to follow, according to the company. Optical input and output is leaving the laboratory and entering the rack.


Why the Photonic Computing Market Is Entering a New Phase


Three structural forces are converging.


AI clusters are power and bandwidth limited. SRI's research identifies AI cluster power and bandwidth limits as the largest driver, adding 4.5 percentage points a year to growth, and the energy cost of AI inference a further 3.0 points. Every generation of accelerator needs more data moved per watt, and copper cannot carry it far enough. The same limit is driving the market for 800G optical transceivers in AI data centres and the shift to liquid cooling for AI racks.


Co-packaged optics is becoming a product. SRI's research finds that co-packaged optics adoption adds 3.6 percentage points and that optical interconnects hold 38.00 percent of value. NVIDIA's Quantum-X Photonics switches shipping in 2026 and Ayar Labs' optical input and output chiplets, now part of NVIDIA's NVLink Fusion programme, put light directly beside the processor. That packaging shift is the same one described in how chiplets and advanced packaging are reshaping AI compute.


Foundries are offering photonics as a platform. SRI's research finds that foundry photonics platforms add 2.2 percentage points. When a silicon photonics process is available from a foundry, start-ups can design optical chips without building fabs, and the USD 500 million round for Ayar Labs in March 2026 is a bet that those designs will ship in volume.


These forces explain SRI's forecast of a market that grows sixteenfold between 2025 and 2035, with value concentrating in interconnects today and processors tomorrow, in AI applications and in North America and Asia Pacific.


Photonic computing market data snapshot: USD 550.00 million in 2025 to USD 8.83 billion by 2035 at 32.0 percent CAGR, with Ayar Labs' USD 500 million Series E at a USD 3.75 billion valuation and Lightmatter's USD 4.4 billion valuation

Five Lessons From the Companies Moving First


The start-ups, chip majors and hyperscalers turning photonics from research into product share a recognisable pattern. We have distilled it into five lessons, each followed by a takeaway for a specific leadership function.


Lesson 1: Interconnect First, Compute Second


SRI's research finds that optical interconnects are the largest component at 38.00 percent of 2025 value and grow at 32.8 percent, while photonic processors hold 24.00 percent and grow fastest at 34.3 percent. The companies moving first, Ayar Labs with optical input and output chiplets, Lightmatter with its Passage photonic interposer and NVIDIA with co-packaged switches, are selling light as a link before they sell light as a calculator. The link has a clear customer and a clear economic case: more bandwidth per watt between chips that already exist.


Photonic processors, which perform matrix operations optically, face the restraint SRI's research names as analog precision and noise, removing 3.1 percentage points a year from growth. The interconnect business funds the longer road to optical compute.


"Nobody has to believe in optical computing to buy an optical interconnect. The link pays for itself today, and it pays for the research that comes next."

Takeaway on strategy. Sequence the roadmap: lead with optical interconnect and co-packaged optics products that plug into existing accelerators and switches, and use that revenue and customer access to develop photonic processors. Sell to the bandwidth problem the customer has now.


Lesson 2: Hybrid Architectures Win the Decade


SRI's research finds that hybrid electro-optical architectures hold 42.00 percent of 2025 value and grow at 30.5 percent, that optical tensor processors grow at 34.0 percent and spiking photonic neural networks fastest at 35.5 percent from a 9.00 percent base. The practical path is to use light where it is best, moving data and performing dense linear operations, and electronics where precision, memory and control are needed.


The companies moving first design systems in which photonics sits beside the electronic processor and memory rather than replacing them, which is also how high-bandwidth memory is being packaged beside AI accelerators.


Takeaway on product. Design for hybrid systems: optical links and optical linear algebra paired with electronic control, memory and nonlinear functions. Make the photonic part a chiplet or interposer that fits existing packaging flows rather than a standalone computer.


Lesson 3: Packaging and Lasers Are the Manufacturing Bottleneck


SRI's research names packaging and software maturity as a restraint removing 1.8 percentage points a year, and packaging and laser yield as a challenge. Lasers and modulators hold 18.00 percent of value, and attaching lasers, aligning fibres and integrating photonics with electronic dies at volume is harder than designing the chip. Ayar Labs' work on a rack-scale co-packaged optics reference system with a hardware partner, and Lightmatter's co-packaged switch and compute trays shown in 2026, are attempts to industrialise that packaging.


The companies moving first partner with foundries, packaging houses and system builders early, because the restraint is in the back end of the line, not the front.


"The optical chip is the easy part. Getting a laser attached, a fibre aligned and a yield that a hyperscaler will accept is where photonic companies are made or broken."

Takeaway on manufacturing. Invest in packaging, laser integration and test as core competencies, and lock in foundry and packaging partners before volume ramps. Build reference systems with server and switch makers so that the packaging problem is solved once and reused.


Lesson 4: Software Decides Whether Light Gets Used


SRI's research finds that software and compilers hold 11.00 percent of 2025 value and grow at 33.8 percent, faster than the market, and names software maturity as part of the restraint. A photonic accelerator that cannot be programmed through the frameworks engineers already use will not be deployed, however efficient it is. Ayar Labs' entry into NVIDIA's NVLink Fusion programme is a recognition that fitting into the dominant ecosystem is as important as the optics.


The companies moving first build compilers, drivers and framework integrations alongside the hardware, and they expose photonics through standard interfaces so that customers do not need to rewrite models.


Takeaway on software and ecosystem. Treat compilers, drivers and framework integration as half the product. Join the interconnect and accelerator ecosystems customers already use, and make photonic capability invisible to the model developer.


Lesson 5: North America Pays, Asia Pacific Builds


SRI's research finds that North America holds 40.00 percent of 2025 value, or USD 220.00 million, growing at 30.5 percent, that Asia Pacific holds 28.00 percent and grows fastest of the large regions at 34.0 percent to USD 2.87 billion by 2035, and that Middle East and Africa grows at 33.5 percent on sovereign AI investment. North America has the hyperscalers and the start-ups; Asia Pacific has the foundries, the packaging capacity and a rapidly growing base of AI data centres.


The capabilities that decide who captures this demand are specific:


  • Optical interconnect and co-packaged optics products that plug into existing accelerators and switches.

  • Hybrid electro-optical architectures delivered as chiplets and interposers.

  • Packaging, laser integration and test capability at volume.

  • Compilers, drivers and framework integration within the dominant ecosystems.

  • Foundry and packaging partnerships in Asia Pacific and customer relationships in North America.


Companies evaluating these competitive dynamics in greater detail can use SRI's Photonic Computing Market analysis to assess component, architecture, application and regional segments and the positioning of the leading start-ups, chip majors and system vendors.


Five lessons framework for photonic computing: interconnect first and compute second, hybrid architectures win the decade, packaging and lasers are the bottleneck, software decides whether light gets used, North America pays and Asia Pacific builds

Takeaway on market development. Win design slots with North American hyperscalers and accelerator vendors, and build manufacturing and packaging partnerships in Asia Pacific where silicon photonics capacity is concentrated. Watch sovereign AI programmes in the Middle East as a third customer base.


Where the Opportunity Is Concentrating


Value is moving toward interconnects now and processors later, toward hybrid architectures, AI workloads and Asia Pacific manufacturing.


By component, optical interconnects hold 38.00 percent, and photonic processors grow fastest at 34.3 percent.


By architecture, hybrid electro-optical holds 42.00 percent, and spiking photonic neural networks grow fastest at 35.5 percent.


By application, AI training and inference is both the largest and the fastest growing at 36.00 percent share and 34.5 percent growth, followed by data centre acceleration at 29.00 percent.


By geography, North America holds 40.00 percent, and Asia Pacific grows fastest of the large regions at 34.0 percent.


At the Horizon: What to Watch Over the Next Three Years


Co-packaged optics volume. NVIDIA's Quantum-X Photonics switches shipping in 2026 will show whether co-packaged optics moves from first deployments to standard practice.


Start-up production ramps. Whether Ayar Labs converts its March 2026 USD 500 million round, and Lightmatter its 2024 USD 400 million round, into high-volume shipments will set the pace of the interconnect segment.


Photonic processor accuracy. Demonstrations of optical tensor processors matching electronic precision on production AI models would unlock the fastest growing component in SRI's research.


"The photonic computing market will be built by the companies that treat light as a link first and a calculator second, and that solve the packaging problem before they promise the revolution."

The Strategic Revenue Opportunity


For photonics start-ups, the task is to ship interconnects at volume and fund the longer road to optical compute. For chip and switch majors, the opportunity lies in integrating co-packaged optics into existing platforms before start-ups set the standard. For foundries and packaging houses, the priority is photonics process and assembly capacity. For hyperscalers, the opportunity is to specify optical interconnect early and capture the energy savings SRI's research identifies as a driver. For investors, the signal is the move from valuation rounds to shipment volumes.


Market intelligence turns these questions into decisions. Segment-level analysis shows where component, architecture and application value is concentrating, competitive analysis shows how start-ups and majors are positioned, and regional analysis shows where hyperscaler demand and foundry capacity will pull growth first. This is the path from market signals to strategic insight to revenue impact.


Takeaway on revenue. Treat the photonic computing market as four revenue pools: optical interconnects and co-packaged optics sold on bandwidth per watt, photonic processors sold on inference energy, lasers and modulators sold as components on yield and cost, and software and compilers sold as the layer that makes light usable. Each needs its own pricing, partner and roadmap.


The Next Leg: Preparing for the Journey Ahead


The defining shift in photonic computing is that light has become a necessity for AI infrastructure rather than a research curiosity. That changes who buys photonics, how it is packaged and which companies control the interface between chips. The companies moving first are not the ones promising an all-optical computer; they are the ones shipping interconnects, designing hybrid systems, industrialising packaging, writing the software and partnering across the Pacific.


The next three years will decide which companies own the optical layer of AI compute. Decision-makers should invest in co-packaged optics, hybrid architectures, packaging and software, and track switch shipments, start-up production ramps and processor accuracy. Companies evaluating growth opportunities in this market can use Strategic Revenue Insights' Photonic Computing Market report to assess market size, competitive dynamics, regional opportunities and strategic growth priorities.


Sources: Strategic Revenue Insights, Photonic Computing Market report (2026); Optics and Photonics News, Ayar Labs Series E (March 2026); Lightmatter, Series D announcement (October 2024); NVIDIA, Quantum-X and Spectrum-X Photonics (2026).


Frequently Asked Questions


How big is the photonic computing market?


Strategic Revenue Insights values the global photonic computing market at USD 550.00 million in 2025 and forecasts USD 8.83 billion by 2035, a 32.0 percent compound annual growth rate and a sixteenfold expansion. North America holds 40.00 percent of 2025 value, and optical interconnects are the largest component at 38.00 percent.


What is driving photonic computing growth?


According to Strategic Revenue Insights, AI cluster power and bandwidth limits add 4.5 percentage points a year to growth, co-packaged optics adoption adds 3.6 points, the energy cost of AI inference adds 3.0 points and foundry photonics platforms add 2.2 points. Analog precision and noise is the largest restraint, removing 3.1 points.


Which photonic computing segment grows fastest?


Photonic processors grow fastest by component at 34.3 percent, spiking photonic neural networks fastest by architecture at 35.5 percent, and AI training and inference is both the largest and fastest growing application at 34.5 percent, according to Strategic Revenue Insights. Asia Pacific is the fastest growing large region at 34.0 percent.


Who leads the photonic computing market?


Strategic Revenue Insights estimates Lightmatter leads with 11.5 percent share, followed by Ayar Labs at 8.5 percent, Celestial AI at 7.0 percent, Intel at 6.5 percent and NVIDIA at 6.0 percent. Ayar Labs raised USD 500 million in March 2026 at a USD 3.75 billion valuation, and Lightmatter raised USD 400 million in October 2024 at USD 4.4 billion.



About the authors: This article was prepared by the Strategic Revenue Insights research team, which combines market intelligence, competitive analysis and strategic advisory for growth-stage, mid-market and enterprise organisations across the ICT, semiconductor and electronics sector.


Sector analyst: Lukas Schmidt, Technology and Infrastructure Analyst. Our market estimates combine primary research (interviews, surveys and expert panels) with secondary data (industry reports, databases and public disclosures); read more about how Strategic Revenue Insights works. To discuss this market with our analysts, contact our team.


Comments


Commenting on this post isn't available anymore. Contact the site owner for more info.
bottom of page