Opticore Raises Additional $7.5M In Funding

Opticore, a photonic computing startup founded in 2023, secured $7.5 million as an extension to its seed round, bringing the total funding to $14.5 million. The funding aims to accelerate development of Opticore’s optical processing units (OPUs), which promise 100x energy efficiency for AI computing in data centers, addressing memory bottlenecks in traditional electronics.

Opticore develops photonic chips that integrate optical waveguides and photoelectric multiplication to perform computations with significantly lower energy use than CMOS-based processors. The company’s OPUs handle memory-intensive AI tasks by converting data to optical signals for on-chip movement, enabling O(N²) performance with O(N) devices. This addresses key limitations like the “memory wall” caused by electronic wires’ capacitive losses, which cap clock speeds and increase thermal dissipation. Opticore’s chips are fabricated using standard foundry processes, with co-packaged optoelectronics and high-bandwidth memory (HBM) integration.

Founded by Zaijun Chen (CEO), Mengjie Yu, and Ryan Hamerly, Opticore emerged from academic research at MIT and other institutions. Key milestones include a 2019 theoretical proposal in Physical Review X, a 2023 demonstration of 100x energy efficiency in Nature Photonics, and a 2025 validation in Science Advances. The company is based in Fremont, California, and focuses on datacenter applications for large language models and other AI workloads.

Funding Details

Opticore’s funding history consists of two related rounds: an initial seed in late 2024 and a 2025 extension.

• Initial Seed Round (December 2024): Raised $7 million. Co-led by Jetha Global (Karan Danthi) and Origin Ventures (Prashant Shukla), with participation from Sagax Capital (Alex Turnbull) and Bioeconomy.XYZ. The funds supported the company’s launch, prototype scaling, and initial chip fabrication. This round marked Opticore’s official emergence from stealth, emphasizing its patented technology for optical GPUs.
• Seed Extension (September 2025): Raised an additional $7.5 million, increasing total funding to $14.5 million. Led by returning investors Jetha Global and Origin Ventures, the extension focuses on advancing photonic chip prototypes toward commercial viability. No new lead investors were named, but the involvement of deep-tech specialists underscores confidence in Opticore’s approach. The announcement highlighted partnerships for silicon photonics scaling and optoelectronic packaging.

Total funding to date: $14.5 million across these rounds. No prior pre-seed or other stages are documented publicly.

Investors and Strategic Implications

The investor syndicate comprises deep-tech focused firms with expertise in hardware and AI infrastructure:

These backers provide not just capital but strategic guidance, leveraging networks in foundries and AI ecosystems. For instance, advisors like MIT’s Dirk Englund connect Opticore to academic talent and validation. The funding aligns with broader trends in photonic computing, where investors seek alternatives to power-hungry GPUs amid AI’s energy crisis—data centers could consume 8% of global electricity by 2030. Opticore’s backers position it to compete with players like Ayar Labs (recently raised $155 million Series D with Nvidia participation), potentially enabling co-packaged optics for hyperscalers.

Technology and Market Context

Opticore’s core innovation is temporal multiplexing in photonic logic, encoding neural data via optical pulses to process billions of parameters on a single chip. This overcomes size constraints of photonic devices (larger than transistors) by requiring fewer components per computation. Demonstrated efficiencies could reduce AI training energy by 100x, critical as models like GPT-4 already demand massive power.

In the market, photonic computing is gaining traction for AI acceleration. Competitors include Lightmatter (raised $400+ million) and Lightelligence, focusing on optical interconnects. Opticore differentiates with integrated OPUs for end-to-end processing. The 2025 extension funds testing larger chips and HBM hyperbonding, targeting deployment in 2026-2027. Challenges include fabrication yields and ecosystem adoption, but DARPA awards (e.g., 2023 NaPSAC, 2024 INSPIRED) and SPIE/Optica recognitions bolster credibility.

With $14.5 million secured, Opticore plans to prototype multi-chip systems and engage data center partners. Success could enable “leapfrog” AI models unattainable with electronics alone, per advisor Dirk Englund. However, the field faces hurdles like optical-electronic conversion losses and supply chain dependencies. If validated, Opticore may pursue a Series A in 2026, potentially valuing it at $100-200 million based on comparable photonic startups. The funding reflects investor optimism in photonics as a sustainable AI enabler, though broader market volatility in deep tech could influence timelines.

Recommended: Inside Rostovsky Watches: Rare Timepieces & Collector Expertise in Beverly Hills

Opticore’s latest funding extension represents a pivotal step in bridging photonic research toward practical AI hardware, amid escalating demands for efficient computing. Founded in 2023 by experts from MIT and Stanford, the company has rapidly progressed from theoretical foundations to tangible prototypes, leveraging exclusive patents on photoelectric multiplication and coherent detection for photonic logic. This technology fundamentally reimagines data movement and computation: traditional electronic systems suffer from the “memory wall,” where shuttling data via metallic wires incurs high capacitive losses, limiting speeds to 1-3 GS/s and driving up energy use. Opticore’s optical processing units (OPUs) mitigate this by converting high-bandwidth memory (HBM) data into optical beams, propagated via low-loss waveguides for on-chip operations. This enables quadratic scaling—O(N²) computations with linear device counts—while consuming 100x less energy than equivalent CMOS setups.

The initial seed round, closed on December 19, 2024, for $7 million, catalyzed Opticore’s launch from stealth mode. Co-led by Jetha Global and Origin Ventures, it drew from a cadre of deep-tech investors attuned to hardware innovations. Karan Danthi of Jetha Global, with a track record in semiconductors, and Prashant Shukla of Origin Ventures, who champions grid-resilient tech, provided not only capital but also strategic insights into scaling. Sagax Capital’s Alex Turnbull, a veteran in multi-asset investments, and Bioeconomy.XYZ, focused on AI-bio intersections, rounded out the syndicate. Some reports cite $5 million, possibly reflecting initial commitments before full close, but consensus points to $7 million enabling early fabrication runs and team expansion in Fremont.

Building on this momentum, extension injected $7.5 million, elevating total capital to $14.5 million. Again co-led by Jetha and Origin, the round emphasizes prototype advancement: larger dies integrating silicon photonics, optoelectronic co-packaging, and HBM via hyperbonding. Shukla noted in the announcement that electronic AI’s energy intensity is “straining electrical grids,” positioning Opticore’s photonics as a core solution for datacenter scale-out. No new investors joined, signaling strong existing backer confidence, though undisclosed follow-ons from seed participants may contribute.

To contextualize, Opticore’s funding trajectory mirrors the photonic sector’s surge. Global AI compute power is projected to multiply 100x by 2027, per industry estimates, but electronic limits—exacerbated by Moore’s Law slowdown and Dennard scaling breakdown—necessitate alternatives. Photonic firms like Ayar Labs ($155 million Series D in 2024, backed by Nvidia) target optical I/O, while Lightmatter’s $400+ million fuels full photonic accelerators. Opticore carves a niche with integrated OPUs for memory-bound tasks like transformer inference, validated through milestones: a 2019 Physical Review X theory by co-founder Hamerly and advisor Englund; 2023 Nature Photonics demo of 100x efficiency; and 2025 Science Advances on scalable logic. DARPA grants (NaPSAC 2023, INSPIRED 2024) and awards (SPIE AI/ML Best Paper 2023, Optica Foundation 2023) further de-risk the tech.

Investor profiles underscore strategic alignment:

These backers, combined with advisors like Englund (MIT Quantum Photonics Lab), facilitate access to foundries (e.g., TSMC-compatible processes) and hyperscalers. Englund’s comment on “leapfrogging” unreachable models highlights the tech’s ambition: processing billions of parameters per chip, dynamically programmable for training/inference.

Market dynamics amplify the round’s significance. AI datacenters could draw 1,000 TWh annually by 2026—equivalent to Japan’s total electricity—prompting innovations like Opticore’s. Competitors abound: PsiQuantum chases fault-tolerant photonics for quantum-AI hybrids; Xanadu builds optical quantum chips. Yet Opticore’s classical focus (no quantum overhead) suits near-term AI needs. Challenges persist: photonic devices’ size demands clever multiplexing (Opticore’s temporal encoding pulses data over time), and hybrid electro-optic interfaces must minimize losses. Fabrication yields, currently a bottleneck, improve with the extension’s funds for validation.

Looking ahead, Opticore eyes 2026 pilots with cloud providers, potentially partnering on co-packaged optics like Sivers Semiconductors’ laser arrays. Total funding of $14.5 million positions it for a Series A, where valuations for similar firms range $100-500 million (e.g., Lightelligence at $300 million post-seed). Success hinges on demos proving 100x gains at scale, amid a frothy AI investment landscape—$100+ billion flowed to AI startups in 2024-2025. If photonic adoption accelerates, Opticore could redefine datacenter economics, enabling sustainable AI proliferation. Conversely, integration hurdles or electronic improvements (e.g., 3nm nodes) might temper hype, but current traction suggests a promising trajectory.

Please email us your feedback and news tips at hello(at)superbcrew.com