Fact-checked by Grok 2 weeks ago

Optical computing

Optical computing is a in information processing that utilizes photons and optical phenomena to perform computations, rather than relying solely on electrons as in traditional electronic computing, enabling advantages such as higher parallelism, reduced , and immunity to through the manipulation of light's properties like , , and . This field encompasses both optical computing, which employs via components like optical transistors and to execute discrete operations, and analog optical computing, which leverages continuous signals for tasks such as transforms, , and matrix multiplications using devices including spatial modulators and lenses. Key principles include exploiting light's inherent parallelism—facilitated by techniques like (WDM) and mode-division multiplexing (MDM)—to process multiple data streams simultaneously, as well as the natural ability of optical systems to perform linear operations at speeds up to 10^17 bit operations per second in specialized setups. Historically, optical computing traces its origins to the , when military applications drove interest in optical transforms for image processing, evolving through the 1980s with advancements in and early demonstrations like the self-electro-optic effect device () developed at AT&T Bell Labs for optical switching. Notable milestones include the 2021 IBM–Skoltech optical switch achieving 1 trillion calculations per second and, more recently, the 2021 photonic tensor core for accelerating matrix operations in applications. Compared to electronic computing, optical systems address critical challenges like the stagnation of beyond 5 nm nodes and the escalating power demands of AI workloads—such as deep neural networks with billions of parameters—by offering lower heat generation, higher bandwidth, and scalability through photonic integrated circuits. For instance, photonic Ising machines, which solve NP-hard optimization problems using optical parametric oscillators, have demonstrated scalability to 2000 nodes in fiber-based systems and 64x64 matrices on chips. As of 2025, while fully optical computers remain in stages due to challenges like the lack of compact optical and transistors, electro-optical architectures are advancing, with commercial prototypes like Lightmatter's 4096 Mach-Zehnder interferometer (MZI) targeting acceleration and Optalysys systems for inference. Recent developments include China's photonic quantum promising 1000-fold gains for complex computing tasks and fully integrated photonic processors for deep s. Future prospects include neuromorphic photonic processors for and , driven by ongoing innovations in metasurfaces and diffractive s.

Fundamentals and History

Definition and Principles

Optical computing is a computational that utilizes photons, or waves, to perform information processing and tasks, in contrast to traditional electronic computing which relies on electrons. This approach exploits the unique properties of , such as wavelength multiplexing for simultaneous data channels, inherent parallelism through of beams, and for pattern-based operations. At its core, optical computing is grounded in the principles of wave optics, including interference, where light waves superimpose to produce constructive or destructive patterns for logical operations; diffraction, which enables the manipulation of light wavefronts for computing tasks like Fourier transforms; and polarization, which allows encoding of information through the orientation of light's electric field vector. Information is typically represented in binary form using photons, where the presence or high intensity of light signifies a '1' bit and absence or low intensity denotes a '0' bit, facilitating direct translation to digital logic. Compared to electronic computing, optical systems offer significant advantages, including higher capable of terahertz-scale data rates due to the vast optical , reduced heat dissipation from the absence of electrical resistance and , and massive parallelism that processes multiple operations simultaneously across spatial or domains. Propagation speed in optical media is governed by the equation for signal delay \tau = \frac{n L}{c}, where L is the path length, n is the of the medium (typically 1.5 for silica), and c = 3 \times 10^8 m/s is the in ; this yields delays on the order of picoseconds per millimeter, far surpassing electron drift velocities (around 10^{-4} m/s in conductors) and enabling lower for long-distance signals without the RC delays common in . However, optical scalability is constrained by challenges like limits, which blur fine features at sub- scales, and the immaturity of all-optical and nonlinear elements.

Historical Development

The concept of optical computing emerged in the with pioneering proposals leveraging light's properties for . Adolf Lohmann introduced computer-generated holograms in 1966, enabling optical transforms for efficient image processing and laying groundwork for analog optical computers. Early ideas also included holographic systems, proposed as high-density alternatives to electronic methods, capitalizing on light's parallelism for data retrieval. These developments focused on analog systems for signal and image processing, marking the initial shift from purely electronic computation. During the 1970s and 1980s, research emphasized practical implementations like optical correlators for and shadow-casting techniques for logic operations. Optical correlators, building on Vander Lugt filters, enabled rapid matched filtering in and applications. Shadow casting, demonstrated in systems using masks and projections to perform logic gates, represented early attempts at discrete optical logic without electronic intermediaries. This era saw limited scalability due to material constraints, but it established foundational architectures for parallel optical operations. The 1990s brought a resurgence driven by advances in semiconductor lasers and , enabling more efficient all-optical . These technologies facilitated demonstrations of optical switching and logic elements, reducing reliance on electro-optic conversions. In the and , with transformed optical computing from discrete components to on-chip systems, compatible with existing semiconductor fabrication. This period saw the development of photonic integrated circuits, with a key 2010 demonstration of a 4x4 non-blocking electro-optic switch matrix on , enabling scalable optical interconnects for computing networks. Efforts at institutions like advanced photonic-electronic chips, addressing bottlenecks in centers. The 2020s accelerated optical computing research amid surging AI demands for energy-efficient, high-speed processing, as electronic systems neared limits. Breakthroughs include Lightmatter's Passage photonic , released in 2023, which uses 3D optical interposers for massive bandwidth in matrix multiplications critical to neural networks. These advancements reflect a from pure optical systems—plagued by lossy nonlinearities—to hybrid electro-optic approaches, where handles movement and while manage control, driven by the impending saturation of scaling. As of 2025, ongoing research continues to focus on scalable photonic processors for AI and beyond.

Core Components and Technologies

Optical Logic Elements

Optical logic elements serve as the foundational building blocks for performing operations in optical computing systems, where light signals encode information through , , or states. These devices exploit optical , nonlinear effects, and switching mechanisms to execute functions without converting to electrical signals, enabling potential advantages in speed and parallelism over electronic counterparts. Key implementations focus on , where '' typically corresponds to low or absent light and '' to high . Core optical switches, such as those based on Mach-Zehnder interferometers (MZIs), are widely used to realize AND and OR gates by inducing controlled phase shifts in the interferometer arms. In an MZI configuration, an input beam is split into two paths by a , one path experiences a via nonlinear interactions, and the beams recombine at a second splitter to produce -dependent output intensity. For instance, a phase shift of 0 radians yields constructive (high output for OR-like behavior), while π radians results in destructive (low output for AND selectivity when both inputs are required). Nonlinear optical materials like enable transistor-like functionality in these switches through its strong electro-optic and nonlinear properties, allowing intensity or akin to gain control in transistors. Specific logic gate designs leverage these principles for precise Boolean operations. The NOT gate is commonly implemented using a continuous-wave probe beam whose polarization is rotated by the input signal via cross-polarization modulation in a nonlinear medium, such as a highly nonlinear fiber or . When the input is '0' (absent or low), the probe passes through an analyzer (output '1'); when input is '1', the rotation (typically 90 degrees) blocks the probe (output '0'). This design operates ultrafast, with switching times limited by the nonlinear response. The for the NOT gate is:
Input (A)Output (NOT A)
01
10
The AND gate employs -based intensity modulation, often in an MZI or diffractive structure, where two input beams must both be present to achieve constructive and produce a high-intensity output; a single input results in partial or no , yielding low output. In this setup, inputs are phase-encoded and combined such that their vector sum modulates the output intensity proportionally to the product of inputs. The for the AND gate is:
Input (A)Input (B)Output (A AND B)
000
010
100
111
This approach provides high contrast ratios, with experimental demonstrations achieving over 20 extinction at multi-gigabit rates. The utilizes a Sagnac loop interferometer, typically with a (SOA) at the center to induce cross- modulation between counter-propagating signals from the two inputs. The loop's symmetry is broken by the nonlinear shift: when inputs differ, the mismatch directs to the output (high intensity), while identical inputs restore balance and suppress output. This configuration supports ultrafast operation up to 40 Gb/s with low . The for the is:
Input (A)Input (B)Output (A XOR B)
000
011
101
110
A critical aspect of MZI-based logic elements is the phase shift in the interferometer arm, governed by the equation \Delta \phi = \frac{2\pi}{\lambda} \Delta n L where \lambda is the operating , \Delta n is the change induced by the input signal via nonlinear or electro-optic effects, and L is the interaction length along the waveguide. This derives from the wave phase accrual, \phi = (2\pi / \lambda) n L, with \Delta \phi representing the differential shift relative to the unmodulated arm; for logic operations, \Delta \phi = \pi typically switches the output from high to low by inverting the condition. Simulations of these elements require modeling the full , including , , and nonlinear Kerr coefficients, often using finite-difference time-domain (FDTD) methods or approaches to evaluate switching efficiency, (typically <3 dB), and phase stability under varying input powers. Such models confirm that \Delta n values on the order of $10^{-4} suffice for \pi-shifts in cm-scale devices at \lambda \approx 1550 nm. In binary digital optical computing, these elements replicate electronic gate behavior by mapping light intensity to logic levels but capitalize on inherent optical parallelism: arrays of MZIs or Sagnac loops can process multiple bits concurrently via in free space or multimode , or across channels. For example, a single multimode waveguide engine can execute operations on 2-bit parallel inputs, achieving throughput scaling with the number of modes while maintaining low . This parallelism addresses bandwidth bottlenecks in electronic systems, with demonstrated at tens of Gb/s per bit stream.

Photonic Integration and Materials

Photonic integration is pivotal for realizing compact and scalable optical , primarily through the of photonic integrated circuits (PICs) that combine multiple optical components on a single chip. serves as a foundational material platform due to its compatibility with complementary metal-oxide- () processes, enabling hybrid electro-optic chips that leverage existing fabrication infrastructure for cost-effective production. This compatibility allows for the monolithic integration of photonic and electronic elements, facilitating seamless data transfer between optical and electrical domains in applications. III-V semiconductors, particularly (InP), are essential for on-chip light sources such as lasers, which are integrated heterogeneously onto substrates to overcome the limitations of indirect-bandgap . These materials provide high-gain active components, enabling efficient light generation and amplification critical for optical . Polymers and metamaterials further enhance integration by offering tunable nonlinear optical effects, such as and Kerr nonlinearity, which are vital for all-optical switching and computation without electronic conversion. PICs are fabricated using wafer-scale techniques, including and , to produce dense arrays of and interconnects on substrates like silicon-on-insulator (SOI). Common waveguide structures include and types: waveguides provide strong light confinement with partial for lower losses, while waveguides offer full for higher index contrast and tighter bends. Coupling between optical fibers and PICs is achieved via grating couplers, which can exhibit losses below 1 dB through optimized and directional coupling designs, minimizing reflections and modal mismatch. Scaling photonic integration faces significant challenges, including dispersion management to maintain signal integrity over varying wavelengths and loss minimization to preserve computational efficiency. In silicon waveguides operating at 1550 nm, typical propagation losses range from 1 to 2 dB/cm, primarily due to sidewall roughness and material absorption, necessitating advanced polishing and cladding optimization. Recent advances in 2024 and 2025 have focused on 3D integration through stacked photonic layers, enabling vertical interconnects that increase density without expanding footprint, as demonstrated in multimaterial platforms combining silicon nitride and lithium niobate. Plasmonics has also progressed, utilizing metal-dielectric structures for sub-wavelength light confinement below the diffraction limit, enhancing nonlinear interactions for compact optical processors while addressing thermal management.

Computing Architectures

Digital Optical Systems

Digital optical systems aim to emulate digital computers using for processing, leveraging optical to perform arithmetic and logical operations in a structured, transistor-like manner. These systems typically incorporate an (ALU) and memory elements constructed from cascaded optical , such as AND, OR, and NOT operations realized through nonlinear optical effects in materials like or s. For instance, an all-optical ALU based on micro-ring resonators can perform addition and comparison functions by cascading basic , achieving operations at telecommunication wavelengths around 1550 nm. Similarly, memory units, such as all-optical D flip-flops, utilize quasi-square ring resonators and T-splitters in platforms to store states, enabling with stabilization times under 1 and footprints of approximately 250 μm² (21 μm × 12 μm). Clocking in these systems is often achieved via pulsed lasers, which provide for high-speed operations by generating ultrafast optical pulses that align activations and , mitigating timing in cascaded networks. Architectures in digital optical systems frequently adopt a von Neumann-style design, featuring separated and memory units interconnected by optical buses for data transfer. These buses exploit to parallelize bit streams, reducing latency compared to electronic counterparts while maintaining binary compatibility. A notable early example from the 1980s-1990s is the shadow-casting system, which used encoded binary images and spatial light modulators to perform parallel addition through overlapping light projections, enabling multi-bit operations like 4-bit addition in a single step via simple decoding masks. This approach demonstrated the potential for digital but was limited by alignment precision and scaling to larger word lengths. Performance in modern prototypes highlights clock speeds exceeding 100 GHz, as demonstrated in an all-optical that integrates linear operations, nonlinear functions, and without electronic conversion, enabling tasks like waveform at rates far beyond electronic limits of ~5 GHz. However, challenges include bit error rates influenced by and contrast ratios, often requiring thresholds optimized to achieve error probabilities below 10^{-9} for reliable multi-stage operation. limits further constrain scalability, governed by for the maximum number of gates a signal can drive: F = \frac{P_{\text{in}}}{P_{\text{min}} \cdot N} where P_{\text{in}} is the input optical power, P_{\text{min}} is the minimum detectable power per gate (typically limited by noise floors around -30 dBm), and N represents the number of cascaded stages; this yields practical fan-out values of 4-8 before amplification is needed, trading off power efficiency against error accumulation. The viability of purely digital optical systems for general-purpose computing remains debated, primarily due to nonlinearity thresholds in optical materials, which demand high intensities for reliable switching, leading to thermal issues and energy inefficiencies in cascaded designs. This has prompted a preference for photonic-electronic approaches, where handle high-speed interconnects and manage precise nonlinearity, balancing performance gains with practical integration.

Analog and Hybrid Approaches

Analog optical computing leverages continuous-valued signals to perform computations, bypassing the constraints of systems and enabling efficient processing of linear operations central to tasks like . In optical s (ONNs), multiplications are realized using spatial modulators (SLMs) to encode weights, where incoming patterns representing input vectors interact with the modulated phase or amplitude to produce output via or . For instance, a 2025 demonstration integrated an SLM to store weights, achieving parallel multiplications for with sub-picosecond latencies. This approach exploits the parallelism inherent in propagation, allowing thousands of operations to occur simultaneously without sequential clocking. Fourier optics further enhances analog capabilities by accelerating convolutions, a core operation in convolutional neural networks (CNNs), through optical fast Fourier transforms (FFTs). Light passes through a to transform the input into the frequency domain, where filtering via masks or modulators applies the convolution kernel, followed by an inverse transform to yield the output. This method has been implemented in platforms, enabling on-chip CNN acceleration with reduced footprint compared to electronic counterparts. A 2024 system using diffractive optics for parallel convolutions demonstrated direct feature extraction in strongly environments, highlighting its robustness for tasks. Hybrid approaches combine optical analog processing with electronic components to mitigate limitations in programmability and non-linear operations, using electro-optic interfaces such as modulators for input encoding and photodetectors for output readout. These systems interface photonic layers for high-speed linear computations with electronic control for activation functions and training updates. A notable 2025 advancement involves silicon photonic-electronic co-processors tailored for edge AI, integrating CMOS-compatible modulators and detectors to achieve seamless data flow in resource-constrained devices. Such hybrids benefit from optical handling of matrix-vector multiplications (MVMs), where the computation time is propagation-limited: for a device of length L, the latency scales as t \approx \frac{L}{n c_0}, with n the refractive index and c_0 the speed of light in vacuum; this derives from the wavefront traversal time across the photonic array, contrasting with electronic delays from RC charging. The energy efficiency of these analog and hybrid paradigms stems from passive optical linear operations, which avoid the dynamic power dissipation of electronic transistors during MVMs—operations that dominate AI workloads. Optical implementations can achieve up to 10 times lower power for matrix operations compared to digital electronic systems, as light propagation incurs negligible energy loss for interference-based multiplications, limited primarily by modulator overhead. For example, photonic tensor cores have demonstrated up to 6× energy reduction relative to electronic accelerators while maintaining high throughput. A seminal application is volume holographic storage for associative memory, pioneered in the 1980s using thick photorefractive crystals to record page-oriented data via angle multiplexing, enabling content-addressable recall through partial input illumination. This technique, developed by Psaltis and colleagues, supports error-tolerant pattern recognition and has influenced hybrid systems into the 2020s for optical data processing.

Advanced and Unconventional Techniques

Time-Delay and

Time-delay methods in optical computing leverage temporal through feedback loops to perform computations, utilizing integrated delay lines coupled with nonlinear optical nodes to generate complex, high-dimensional state spaces. These systems often operate at the edge-of-chaos , where the balance and to inputs, enabling rich nonlinear transformations suitable for processing time-dependent data. A common nonlinear node is the semiconductor optical amplifier (SOA), which provides intensity-dependent gain and phase shifts to map input signals into evolving states. Reservoir computing (RC) adapts these time-delay architectures to optical implementations, where a delayed loop in a nonlinear optical system acts as the fixed reservoir, and only the linear readout layer is trained. Pioneered in all-optical setups using off-the-shelf telecom components, such systems process inputs via optoelectronic and through delay lines, with the nonlinearity arising from devices like integrated lasers or SOAs. From 2017 onward, evolutions have focused on enhancing speed and scalability, including hybrid optoelectronic designs achieving at rates up to 10 Gbps with low error rates, and recent 2024-2025 advances incorporating distributed in fibers or for improved energy efficiency and task performance on chaotic . The principles of optical RC draw from echo state networks, where the reservoir's recurrent dynamics project inputs into a high-dimensional space without altering the internal weights. In time-delay optical mappings, the reservoir state evolves according to a approximating the discrete update: \mathbf{x}(t + \tau) = f(\mathbf{W} \mathbf{x}(t) + \mathbf{W}_{in} u(t)) Here, \mathbf{x}(t) is the , \tau is the delay time, f is a nonlinear (optically realized via SOA gain saturation or ), \mathbf{W} represents the fixed recurrent weights (emulated by virtual nodes along the delay line), \mathbf{W}_{in} is the input coupling, and u(t) is the input signal. The optical mapping discretizes the delay line into N virtual nodes via time , with states read out as intensity samples. These approaches offer advantages in training-free computation for time-series tasks, as the reservoir's fixed dynamics handle nonlinear feature extraction, requiring only for the output weights, which enables rapid adaptation to tasks like prediction or without .

Wavelength and Spectral Methods

(WDM) enables parallel processing in optical computing by transmitting multiple data signals simultaneously over a single using distinct s of light, allowing independent logic operations on each channel without interference. This approach leverages the domain to achieve high parallelism, where each wavelength carries a separate computational thread, such as bitwise operations or , processed via photonic circuits. In optical logic applications, WDM structures facilitate scalable architectures by and manipulating signals based on their spectral signatures, outperforming single-wavelength systems in throughput. Arrayed waveguide gratings (AWGs) serve as key components for wavelength routing in WDM-based optical computing, demultiplexing input signals into individual wavelength channels and recombining them after processing. These integrated photonic devices consist of an input , a free propagation region, an array of waveguides with incrementally increasing path lengths, and an output slab, enabling dispersive routing with low . The channel spacing \Delta \lambda in an AWG is determined by the formula \Delta \lambda = \frac{\lambda_0^2}{n_c N \Delta L}, where \lambda_0 is the central wavelength, n_c is the effective group index of the arrayed waveguides, N is the number of channels, and \Delta L is the path length difference between adjacent waveguides. This design allows precise separation of wavelengths, supporting dense multiplexing for logic gates and arithmetic units in photonic processors. Spectral holography extends WDM principles to volume storage and computation by recording multiple data pages at different wavelengths within the same holographic medium, enabling parallel readout and associative processing. In computing applications, holographic associative memories use wavelength-selective interference to perform pattern recognition or vector-matrix multiplications, where spectral shifts address specific holograms for logic evaluation. This technique combines storage density exceeding 1 Tb/cm³ with computational capabilities, as angle and wavelength multiplexing allow superimposed operations without physical separation. Coherent spectral computing exploits phase-coherent multi-wavelength signals for parallel arithmetic, as demonstrated in a 2023 photonic using WDM to encode inputs across wavelengths for synaptic operations, achieving error-free on up to channels. A more recent advancement in 2025 integrated a microcomb source with a Mach-Zehnder interferometer to perform 100-wavelength parallel multiplications, validating broadband processing over 40 nm for tasks like . These systems highlight methods' ability to scale to over 1000 parallel channels in dense WDM configurations, surpassing limits constrained by and alignment tolerances in single-wavelength setups. In 2025, integrated spectral processors advanced to telecom speeds for , employing multi-wavelength chiral elastomers to stretch-tune spectra for programmable keys, enabling secure data encoding at rates exceeding 100 Gb/s with resistance to attacks. This development integrates AWG with holographic encoding, providing dual-layer protection via wavelength-specific for real-time cryptographic operations in optical networks.

Spatial and Beam Manipulation Techniques

Spatial and beam manipulation techniques in optical computing exploit the spatial properties of light beams to encode and process information, enabling parallel operations through physical interactions like masking and . One foundational approach is beam masking using spatial light modulators (SLMs), which dynamically control light amplitude or phase to perform logic operations via shadow casting. In this method, input patterns are projected from multiple point sources through coded masks, creating overlapping shadows on a detector that represent logical functions such as , or more complex multi-valued operations. Developed prominently in the , this technique allows direct table look-up for logic without intermediate encoding, addressing limitations in earlier coded systems by using noncoded masks for efficient . A key application of shadow casting involved joint transform correlators, where SLMs facilitated real-time by generating sharper autocorrelation peaks through programmable masking. These systems, demonstrated in the late 1980s, utilized incoherent light and SLM-based input modulation to correlate images, achieving high-speed detection of features in scenes like character recognition or object matching. Earlier precursors trace back to the , with Lohmann's methods for optical automata employing photographic copying—often termed xeroxing—onto transparencies to prepare input patterns for recognition tasks. Lohmann's approach integrated these copied transparencies into coherent optical processors, leveraging transforms for parallel in early information processing systems. Contemporary advancements build on these principles with diffractive deep neural networks (D2NNs), introduced by UCLA researchers in , which stack multiple layers of passive diffractive surfaces to manipulate beam propagation for tasks. Fabricated via low-cost , these all-optical networks perform spatial transformations through light diffraction, classifying images such as handwritten digits (MNIST ) with up to 93% accuracy using 7-10 layers, or enabling unit-magnification at terahertz frequencies. Subsequent developments from 2019 to 2021 extended D2NNs to wavelength processing and ensemble configurations, enhancing classification accuracy for diverse applications while maintaining passive, power-free operation. These techniques rely on diffraction physics for computation, often approximated using the Fresnel model for near-field beam evolution in practical setups. The intensity of the diffraction pattern is described by I(\theta) = \left| \int E(x) \exp(i k x \sin\theta) \, dx \right|^2, where E(x) is the input field, k is the , and \theta is the diffraction angle; this formulation, rooted in the Huygens-Fresnel principle, underpins the passive spatial processing in D2NNs and similar systems, offering scalable, low-cost solutions for vision-based tasks without active electronics.

Optimization and Specialized Processors

Optical computing has advanced through specialized processors tailored to computationally intensive tasks, such as solving NP-hard optimization problems and accelerating matrix operations essential for . These systems leverage the inherent parallelism and to outperform traditional counterparts in targeted domains, enabling efficient handling of problems that are intractable on classical . By mapping complex algorithms onto photonic architectures, researchers have demonstrated practical implementations that achieve substantial performance gains while maintaining low . Ising machines represent a prominent class of optical optimization processors, designed to solve problems by simulating the of interacting spins. In these systems, networks of degenerate optical parametric oscillators (DOPOs) emulate spin interactions, where the of the Ising Hamiltonian corresponds to the optimal solution of the mapped problem. A seminal demonstration in 2016 by NTT researchers showcased a 2000-node coherent Ising machine (CIM) using time-multiplexed DOPO pulses coupled via measurement and , successfully addressing all-to-all spin-spin couplings for large-scale optimization. More recent advancements include scalable photonic CIMs applied to , such as bus route optimization in urban transport networks; a 2025 study utilized a CIM to solve multi-objective routing problems for 308 nodes, partitioning them into communities with faster convergence than classical solvers. These optical Ising machines offer significant speedups for NP-hard problems, with benchmarks showing up to orders-of-magnitude improvements in solution quality and time over digital simulated annealing methods. On-chip photonic tensor cores have emerged as specialized processors for accelerating matrix multiplications in workloads, exploiting to perform parallel vector-matrix operations at light speed. These cores integrate photonic Mach-Zehnder interferometer meshes to execute tensor contractions with minimal latency and power draw, making them ideal for inference and training. Lightmatter's 2024-2025 developments, including the platform with integrated photonic tensor cores, demonstrate improved over electronic GPUs for acceleration in hybrid electro-optic systems. Optical Fourier co-processors provide ultrafast spectrum analysis for applications, realizing the (DFT) through free-space . In a lens-based setup, an input signal modulated onto a spatial is focused by a , where the focal plane naturally computes the via ; the output intensity and distributions yield the frequency-domain representation. Mathematically, this corresponds to the continuous : F(u) = \int_{-\infty}^{\infty} f(x) \exp(-i 2\pi u x) \, dx where f(x) is the spatial input distribution and F(u) is the transform at frequency coordinate u in the focal plane. A 2017 implementation demonstrated a complex-to-complex DFT coprocessor using direct phase determination, enabling real-time processing of terabit-per-second signals with sub-picosecond latencies unattainable by digital FFTs. Such processors are particularly valuable in radar and imaging systems, offering 1000-fold speedups for large-scale transforms compared to electronic counterparts.

Challenges and Future Prospects

Technical and Practical Limitations

One of the primary physical limitations in optical computing stems from the weak intrinsic nonlinearity of optical materials, which hinders efficient photon-photon interactions essential for operations. Unlike electronic systems where transistors enable straightforward nonlinearity at low power levels through effects, optical nonlinearity—primarily via the —requires impractically high light intensities to achieve significant phase shifts, typically on the order of 1 /cm² in materials like fused silica. This threshold arises because the nonlinear refractive index coefficient n_2 for common dielectrics is small, approximately $3 \times 10^{-16} cm²/W, necessitating intense fields to induce measurable changes without damaging components. The nonlinear phase shift is given by \phi_{NL} = \frac{2\pi n_2 I L}{\lambda}, where I is the optical intensity, L is the interaction length, and \lambda is the wavelength; achieving \phi_{NL} \approx 1 radian over practical lengths (e.g., millimeters to centimeters) demands elevated I, limiting scalability and energy efficiency in integrated systems. Crosstalk poses another critical physical barrier, particularly in dense photonic integrations where closely spaced waveguides lead to unintended between channels. In silicon photonic circuits, electro-optic can degrade , with reported levels exceeding -20 dB in high-density arrays, constraining the minimum feature size and overall integration density compared to counterparts. This issue is exacerbated in architectures, where misalignment or fabrication variations amplify , limiting the number of concurrent operations without error correction overhead. On the practical front, fabrication costs for photonic integrated circuits (PICs) remain substantially higher than those for silicon CMOS processes, often by factors of several times due to specialized , lower yields, and the need for precise alignment of optical components. For instance, while CMOS benefits from mature, high-volume production ecosystems, PIC involves complex material deposition and steps that increase per-wafer expenses and reduce throughput. Additionally, power consumption in optical systems is dominated by sources and photodetectors, which can account for a significant portion of overall energy use—lasers alone may require watts per channel, offsetting gains in signal propagation efficiency. Insertion losses in PICs further impede cascadability, with typical fiber-to-chip exceeding 10 dB per facet and on-chip losses adding 1-5 dB per component, resulting in cumulative of 5-10 dB across multi-stage circuits. This restricts the depth of optical logic chains to a few operations before is needed, introducing and . Common misconceptions about optical computing include the notion of "zero heat" generation, which overlooks thermal contributions from active components like modulators and amplifiers, as well as the requirement for cryogenic cooling in advanced detectors such as superconducting nanowire single-photon detectors (SNSPDs) used in quantum-enhanced systems. While photon propagation itself produces minimal , overall system can necessitate cooling solutions near for noise-sensitive elements, shifting thermal management burdens rather than eliminating them. Similarly, outdated views emphasizing raw speed ignore inherent latencies from optical routing and conversion bottlenecks, where effective throughput may not surpass optimized in short-distance computations. In 2025, research in optical computing has increasingly emphasized AI accelerators that leverage photonic integrated circuits (PICs) for high-speed, low-latency processing of neural networks. These systems, such as integrated neuromorphic photonic platforms, enable on-chip acceleration of tasks by exploiting optical nonlinearities for matrix-vector multiplications, offering sub-nanosecond latency and higher throughput compared to electronic counterparts while consuming significantly less power. Hybrid approaches combining quantum photonics with classical optical computing are also gaining traction, particularly for error-corrected quantum operations integrated into photonic chips. Neuromorphic , inspired by brain-like architectures, has seen notable progress through initiatives like the NEUROPULS project, which develops secure photonic accelerators using phase-change materials for reconfigurable synaptic weights, and the program, focusing on programmable photonic networks for high-speed imaging and communications. Scalability efforts are targeting through free-space optics () interconnects, which facilitate massive parallelism in data centers by transmitting terabits per second over optical beams without the bandwidth limitations of copper wiring. Post-2023 advances in scalable PICs, including platforms from IEEE studies, have enabled denser integration of thousands of optical components on a single chip, supporting workloads like training with reduced thermal overhead. Bio-inspired optics, particularly optical s mimicking neural plasticity, have advanced via 2025 Nature publications on optoelectronic memristors and 2D photodetectors that perform in-sensor , offering adaptive visual processing with sub-milliwatt energy per . Prospects for optical computing include transformative energy efficiencies, with photonic AI accelerators projected to deliver up to 10-fold reductions in operational per inference for operations compared to GPU-based systems, potentially slashing global AI energy demands from petawatt-hours to more sustainable levels. Integration with networks is underway through links and flexible optical architectures like those in the EU's FLEX-SCALE project, enabling ultra-low-latency for holographic communications and at speeds. Ethically, these developments promise reduced e-waste via longer-lasting, recyclable photonic hardware and lower s, aligning with goals in and infrastructure. Key milestones include 's 2025 selections for photonic quantum initiatives, such as PsiQuantum's integrated processors; in October 2025, advanced PsiQuantum to the final phase of its US2QC program for utility-scale using photonic integrated processors.

Industry and Applications

Commercial Developments

Lightmatter has emerged as a leading player in photonic computing, raising $400 million in a Series D funding round in October 2024, which quadrupled its valuation to $4.4 billion and supported the development of its photonic interconnect platform. The company launched the chip in April 2025 as a interposer designed to enable high-speed chip-to-chip communications for workloads, addressing limitations in data centers. In April 2025, Lightmatter introduced an updated Envise photonic chip for tasks, leveraging optical to achieve energy-efficient processing for large-scale neural networks. Ayar Labs has advanced in-package optical interconnects, securing $155 million in a Series D funding round in December 2024 with investments from , , and to scale its TeraPHY technology for infrastructure. The company announced a partnership with in September 2025 to integrate co-packaged , enabling higher bandwidth densities for systems. Ayar Labs' solutions target the replacement of with optical links, supporting the commercialization of multi-terabit-per-second data transfer in clusters. PsiQuantum, primarily focused on photonic quantum computing, has also pursued classical photonic components as part of its hybrid approach, raising $1 billion in a Series E round in September 2025 to fund the construction of utility-scale with integrated optical interconnects. The company broke ground on a major facility in in September 2025, aiming to deploy million-qubit photonic processors by 2027 while leveraging for scalable classical . In February 2025, PsiQuantum achieved a breakthrough in high-volume manufacturing for generation and networking, paving the way for commercial systems. Celestial AI has driven commercialization of optical co-processors, announcing adoption of its Photonic Fabric technology by leading hyperscalers in March 2025, following its $175 million Series C funding round in March 2024 and subsequent Series C1 investments to accelerate deployment. The platform, which uses optical interconnects to disaggregate compute and memory in systems, marked an early with initial shipments to operators in late 2024. By August 2025, Celestial AI secured an additional $255 million in Series C1 funding, bringing total investment to $520 million and supporting broader market rollout. Venture capital investments in optical computing startups exceeded $1 billion from 2023 to 2025, with notable rounds including Lightmatter's $400 million, Ayar Labs' $155 million, PsiQuantum's $1 billion, and Celestial AI's combined $520 million across series as of August 2025. The U.S. CHIPS and Science Act has indirectly bolstered these efforts through $52.7 billion in semiconductor funding, including allocations to photonics firms like Coherent for advanced manufacturing capabilities essential to optical interconnects. A key commercialization milestone in 2025 was the rollout of 1 Tbps optical interconnects, exemplified by Avicena's debut of a microLED-based 1 Tbps for data centers and Broadcom's co-packaged achieving over 1 Tbps/mm density for Ethernet switches. These developments signal the transition from prototypes to production-scale optical links, enabling hyperscalers to scale training without electrical bottlenecks.

Real-World Use Cases

Optical computing has found practical applications in and through photonic accelerators, which enable faster training and inference in cloud deployments as of 2025. These devices leverage light-based matrix multiplications to process operations at speeds exceeding 1 GHz, significantly outperforming traditional electronic counterparts in for large-scale workloads. For instance, platforms have been integrated into hardware to handle sustainable computing demands, reducing power consumption while scaling to handle complex models. In , optical signal supports efficient routing in and backhaul networks by minimizing through all-optical manipulations. Photonic processors designed for signal handling can classify and route data streams in nanoseconds, facilitating seamless integration with infrastructure and enabling low-delay connections for high-bandwidth applications. A notable example is the deployment of all-optical metro networks in regions like , where such systems enhance transmission efficiency across urban areas. For sensing and imaging, optical computing enables real-time processing in systems critical for autonomous , allowing rapid environmental mapping and . Photonic integrated circuits perform these transforms optically, processing vast datasets from scans to support navigation decisions at high speeds. In 2024, automotive integrations of frequency-modulated continuous-wave (FMCW) photonic began appearing in production , improving perception accuracy in varied conditions like low light or adverse weather. Beyond these domains, optical computing addresses optimization challenges in via Ising-based photonic machines, which solve allocation problems by mapping them to spin interactions for near-optimal risk-return balances. In , optical correlators accelerate through parallel , aligning genetic sequences at speeds unattainable by electronic methods alone and aiding in genomic analysis for . A key impact of these applications is demonstrated in hybrid optical-electronic systems, where Microsoft's analog optical computer (AOC) achieved substantial power reductions—up to 90% less energy than conventional GPUs—for real-world tasks like MRI image reconstruction in data centers, highlighting the potential for scalable, efficient deployments across industries.

References

  1. [1]
    Optical Computing: Status and Perspectives - PMC - PubMed Central
    Jun 24, 2022 · The review of the status and perspectives shows that optical technology offers incredible developments in computational efficiency.
  2. [2]
    The challenges of modern computing and new opportunities for optics
    Sep 9, 2021 · In general, optical computing can be classified in two different categories: the digital optical computing and the analog optical computing.Moore's Law And The New... · Photonic Ising Machine · Hybrid Computing System
  3. [3]
    The role of optics in computing | Nature Photonics
    We have shown that an optical approach can in principle be used to create a universal processor with functionality comparable to a Turing machine.<|control11|><|separator|>
  4. [4]
    [PDF] USC-SIPI REPORT #107 - Optical Computing and Interconnections
    This report presents an overview of digital optical computing and optical intercon- nections research. The architectural characteristics of digital optical ...
  5. [5]
    [PDF] Optical Computing: A 60-Year Adventure - HAL
    May 30, 2013 · In 1966, Lohmann revolutionized holography by intro- ducing the first computer generated hologram (CGH) [36,. 37] using a cell oriented encoding ...
  6. [6]
    Optical Computing: Past and Future - Optics & Photonics News
    Jun 1, 2016 · Analog processors. The earliest optical computers, in the 1960s, were analog processors performing image and signal processing tasks—the most ...
  7. [7]
    [PDF] Optical Computing - DTIC
    The subject itself, however, can be traced back to the early 1960s, and many ... optical storage and ideas for optically bistable devices. Fourier and ...
  8. [8]
    Optical Computing | (1986) | Publications - SPIE
    The mathematical and physical basis for realizing compound literals via optical shadow-casting is discussed. The relation between various switching algebras ...
  9. [9]
    Lasers, nonlinear optics and optical computers - ADS
    A review is given of the latest state of research in all-optical nonlinear logic switches, amplifiers and memories, and it is shown that - ...Missing: 1990s resurgence semiconductor transistor IBM 1994
  10. [10]
    Silicon electro-optic 4x4 non-blocking switch array for on-chip ...
    Jul 11, 2011 · Silicon 4x4 switch array based on carrier-injection for switching of multiple 40Gbps WDM channels is reported. Operation in all 9 unique ...
  11. [11]
    Optical computing | MIT News | Massachusetts Institute of Technology
    Researchers have produced a working optoelectronic chip that computes electronically but uses light to move information.Missing: 2000s 2010s 4x4
  12. [12]
    Lightmatter Unveils Passage M1000 Photonic Superchip, World's ...
    Mar 31, 2025 · The Passage™ M1000 enables a record-breaking 114 Tbps total optical bandwidth for the most demanding AI infrastructure applications. At more ...
  13. [13]
    Xanadu Announces Aurora, A Universal Photonic Quantum Computer
    Jan 22, 2025 · Xanadu developed Aurora, a universal photonic quantum computer with a modular architecture of interconnected server racks for scalability.Missing: Passage | Show results with:Passage
  14. [14]
    The Impact of the End of Moore's Law on the AI Gold Rush - EE Times
    Sep 12, 2025 · However, challenges with stability, error correction, and scalability mean that pure quantum processors are at least a decade away. Hybrid ...
  15. [15]
    [PDF] Optical Computing: Status and Perspectives - Semantic Scholar
    Jun 24, 2022 · The optical–electrical transformation might severely limit the energy efficiency of the hybrid computing device, making it impossible to ...
  16. [16]
    Design of Optical Logic Gates Using Mach–Zehnder Interferometers ...
    Jul 12, 2022 · The phase shifts are estimated as the ones that provide the higher distance between the outputs corresponding to the logic levels 0 and the 1.
  17. [17]
    New Optical Switch Could Lead to Ultrafast All-Optical Signal ...
    Jul 29, 2022 · First, they chose a crystalline material known as lithium niobate, a combination of niobium, lithium, and oxygen that does not occur in ...
  18. [18]
    Two all-optical logic gates in a single photonic interferometer
    Nov 15, 2015 · Obtaining optical logic gates – OR, XOR, AND and logic functions using asymmetric Mach-Zehnder interferometer based on photonic crystal fiber.
  19. [19]
    Polarization-based all-optical logic controlled-NOT, XOR, and XNOR ...
    Oct 20, 2011 · In this paper, we demonstrate experimentally three polarization-based all-optical logic gates: controlled-NOT, XOR, and XNOR gates.Missing: seminal | Show results with:seminal
  20. [20]
    All‐Optical Logic Gates: Designs, Classification, and Comparison
    Mar 19, 2014 · It is a polarization dependent phenomenon and capable of handling intensity modulation, phase modulation, and frequency shift keying signal.<|separator|>
  21. [21]
    Realization of optical logic gates using on-chip diffractive ... - Nature
    Sep 21, 2022 · Here, we discuss the design of a multi-functional optical logic gate based on an on-chip diffractive optical neural network that can perform AND, NOT and OR ...
  22. [22]
    Optical Logic Gates Excited by a Gauss Vortex Interference Beam ...
    Apr 20, 2023 · We found that spatial self-phase modulation patterns through the MoS 2 dispersions can be modulated by the initial intensity, phase, and topological charge of ...
  23. [23]
    Optics Communications - ScienceDirect.com
    In this paper, we explore the ability of the SOA-assisted Sagnac switch to perform Boolean bitwise XOR logic between optical signals that carry 10 Gb/s data ...
  24. [24]
    Ultrafast all-optical logic operations in a nonlinear Sagnac ...
    A high-contrast XOR XOR ¯ operation over 20 dB can be achieved if sufficient pump power (>1.8 W) or a longer fiber loop (>500 m) is available, as was confirmed ...
  25. [25]
    All-Optical XOR in a Semiconductor Optical Amplifier-Assisted Fiber ...
    Aug 5, 2025 · All-optical Boolean XOR is demonstrated on a high-speed serial bit stream with a three-optical input fiber Sagnac interferometer switch, ...
  26. [26]
    Comprehensive model of MZI-based circuits for photonic computing ...
    Jul 3, 2025 · We present a comprehensive circuit model for Mach-Zehnder interferometer (MZI) based meshed topologies, that is able to accurately predict the behavior of ...
  27. [27]
    Mach-Zehnder Interferometer - an overview | ScienceDirect Topics
    Phase shifts of the two dimensional images due to the change of temperature are calculated by subtracting all phase images from the initial image. The base of ...
  28. [28]
    Programmable Parallel Optical Logic Gates on a Multimode ... - MDPI
    The device can perform logic AND or OR operations for the input signals A and B, each containing two bits of information, in parallel. The input signals, in the ...
  29. [29]
    All-optical logic gate computing for high-speed parallel information ...
    The proposed systems can potentially perform binary logic operations in parallel for many input bits simultaneously. However, they generally have the ...
  30. [30]
  31. [31]
    Roadmapping the next generation of silicon photonics - Nature
    Jan 25, 2024 · We chart the generational trends in silicon photonics technology, drawing parallels from the generational definitions of CMOS technology.
  32. [32]
    III/V-on-Si MQW lasers by using a novel photonic integration method ...
    Oct 9, 2019 · We demonstrate a novel photonic integration method of epitaxial regrowth of III/V on a III/V-on-SOI bonding template to realize heterogeneous lasers on silicon.
  33. [33]
    Full article: Nonlinear optics in metamaterials - Taylor & Francis Online
    Metamaterials open new possibilities for the realization of efficient nonlinear optical interactions such as wavelength conversion and parametric amplification ...Missing: polymers | Show results with:polymers
  34. [34]
    Low loss shallow-ridge silicon waveguides - Optica Publishing Group
    However, the lowest propagation loss reported for these waveguides to date at λ = 1550 nm is 1-2 dB/cm [7, 8], which is not suitable for applications such as ...
  35. [35]
    Optimization of Grating Coupler over Single-Mode Silicon-on ... - MDPI
    Mar 15, 2024 · In this paper, using a deep-learning model combined with an inverse-design process, we achieve <1 dB losses for grating couplers implemented over single-mode ...
  36. [36]
    Silicon‐Integrated Next‐Generation Plasmonic Devices for Energy ...
    Jul 1, 2025 · Silicon (Si)-integrated plasmonics offer a pathway to next-generation, energy-efficient semiconductor applications.
  37. [37]
    Micro-ring resonator based all-optical Arithmetic and Logical Unit
    Here, we propose an all-optical ALU exclusively using Micro-Ring Resonator (MRR) to perform addition and comparison.
  38. [38]
    Design of All-Optical D Flip Flop Memory Unit Based on Photonic ...
    Aug 6, 2024 · This paper proposes a unique configuration for an all-optical D Flip Flop (D-FF) utilizing a quasi-square ring resonator (RR) and T-Splitter.Missing: ALU | Show results with:ALU
  39. [39]
    50GHz optical pulse generation for synchronization of high-speed ...
    Sep 8, 2006 · In this paper, we design an amplitude modulated mode-locked figure eight fiber laser as a 50 GHz optical pulse. In the cavity of high speed ...
  40. [40]
  41. [41]
    Optical parallel logic gates using a shadow-casting system for ...
    Optical parallel logic gates using a shadow-casting system for optical digital computing ... parallel operations of addition or subtraction for two binary ...
  42. [42]
    Bit error rate of optical logic: fan-in, threshold, and contrast
    ### Summary of "Bit Error Rate of Optical Logic: Fan-in, Threshold, and Contrast"
  43. [43]
    Harnessing optical advantages in computing: a review of current ...
    Mar 14, 2024 · This article examines the evolving landscape surrounding optical advantages in computing, focusing on current trends and prospects.Introduction · High-speed signal processing... · Energy efficiency in optical...
  44. [44]
    Role of all-optical neural networks | Phys. Rev. Applied
    Jan 17, 2024 · Here, we consider the viability of all-optical computing as an alternative to electronic and optoelectronic approaches, and attempt to identify ...<|control11|><|separator|>
  45. [45]
    Analog optical computer for AI inference and combinatorial ... - Nature
    Sep 3, 2025 · The spatial light modulator (2) stores neural network weights or optimization problem coefficients, and multiplies them with the incoming light.
  46. [46]
    Optical neural networks: progress and challenges | Light - Nature
    Sep 20, 2024 · Optical neural networks (ONNs) have made a range of research progress in optical computing due to advantages such as sub-nanosecond latency, low heat ...
  47. [47]
    Combining Silicon Photonics and Fourier Optics for Computer Vision
    Dec 19, 2020 · CNNs principally employ the convolution operation, which can be accelerated using the Fourier transform. In this paper, we present an optical ...
  48. [48]
    Memory-less scattering imaging with ultrafast convolutional optical ...
    Jun 14, 2024 · Training this Fourier optics-based, parallel, one-step convolutional ONN with the strong scattering process for direct feature extraction, we ...
  49. [49]
    Photonic matrix multiplication lights up photonic accelerator and ...
    Feb 3, 2022 · An optical computing system based on digital logic can also be built with all-optical or optoelectronic-hybrid logic computing functions.
  50. [50]
    A Dynamically-Operated Photonic Tensor Core for Energy-Efficient ...
    May 31, 2023 · Our comprehensive evaluation demonstrates that DOTA achieves a >4x energy and a >10x latency reduction compared to prior photonic accelerators, ...
  51. [51]
    Storage capacity of holographic associative memories
    John Hong and Demetri Psaltis, "Storage capacity of holographic associative memories," Opt. ... Cross-talk-limited storage capacity of volume holographic memory.
  52. [52]
    Semiconductor Optical Amplifier (SOA)-Driven Reservoir Computing ...
    Jun 18, 2023 · This paper proposes a reservoir computing scheme using a semiconductor optical amplifier (SOA) to improve signal quality in DWDM, achieving ...
  53. [53]
    Parallel Reservoir Computing Using Optical Amplifiers - IEEE Xplore
    Jul 28, 2011 · This paper investigates using optical amplifiers for parallel reservoir computing, a hardware implementation of reservoir computing, and shows ...
  54. [54]
    Do Reservoir Computers Work Best at the Edge of Chaos? - arXiv
    Dec 2, 2020 · A number of researchers have stated that the highest computational capacity for a reservoir computer is at the edge of chaos.Missing: optical | Show results with:optical
  55. [55]
  56. [56]
    Photonic next-generation reservoir computer based on distributed ...
    Jul 2, 2024 · This work demonstrates a new photonic reservoir computing architecture that uses Rayleigh backscattering in an optical fiber as the main computational engine.Iii. RÖssler System · Data Availability · Appendix A: Photonic Ng-Rc...
  57. [57]
    Optical next generation reservoir computing | Light - Nature
    Jul 21, 2025 · The optical NGRC demonstrated in this work delivers more interpretable results by synthesizing features of time-delayed inputs through the ...Missing: seminal | Show results with:seminal
  58. [58]
    [1207.1619] All-optical Reservoir Computing - arXiv
    Jul 6, 2012 · Here we report an all-optical implementation of a Reservoir Computer, made of off-the-shelf components for optical telecommunications.Missing: seminal | Show results with:seminal<|separator|>
  59. [59]
    High-Speed Photonic Reservoir Computing Using a Time-Delay ...
    Feb 6, 2017 · Our approach exhibits state-of-the-art speed on speech recognition tasks, identifying up to one million words per second with very low error ...
  60. [60]
    Emerging opportunities and challenges for the future of reservoir ...
    Mar 6, 2024 · This Perspective intends to elucidate the parallel progress of mathematical theory, algorithm design and experimental realizations of reservoir computing.
  61. [61]
    Exploiting Wavelength Division Multiplexing for Optical Logic ...
    Wavelength division multiplexing (WDM) is widely used in optical communication for enabling multiple signals being processed and transferred independently.Missing: computing | Show results with:computing
  62. [62]
    [PDF] Neuromorphic computing using wavelength- division multiplexing
    Neuromorphic computing uses wavelength-division multiplexing (WDM) in optical neural networks (ONNs) to enhance computing power and energy efficiency. WDM uses ...<|separator|>
  63. [63]
    [PDF] Wavelength-division-multiplexing based electronic photonic network ...
    In this paper, the WDM-based building blocks for optical computing are introduced. The idea of building a WDM-based optical structure to realize various logic ...
  64. [64]
    Arrayed waveguide gratings for wavelength routing - IEEE Xplore
    Arrayed waveguide gratings (AWGs) can perform wavelength routing for a large number of optical channels and provide a high level of functionality on an ...Missing: computing | Show results with:computing
  65. [65]
    Arrayed Waveguide Gratings – AWG - RP Photonics
    An arrayed waveguide grating is a (typically fiber-coupled) device which can separate or combine signals with different wavelengths.
  66. [66]
    Arrayed Waveguide - an overview | ScienceDirect Topics
    An arrayed waveguide grating (AWG) is a generalization of the Mach-Zehnder interferometer. This device is illustrated in Figure 3.24. It consists of two ...
  67. [67]
    [PDF] ULTRA-HIGH DENSITY OPTICAL DATA STORAGE
    The storage capacity of holography can be enhanced further by adding spectral multiplexing. In a crystal of Pr:YSO, we have demonstrated spectral multiplexing ...
  68. [68]
    Architecture for optical computing using holographic associative ...
    A new architecture for optical computing is proposed, based on associative memory which can be conveniently constructed by optical holography.
  69. [69]
    Phase retrieval in holographic data storage by expanded spectrum ...
    Nov 2, 2023 · The holographic data storage technology records the amplitude and phase of the signal beam in the media in the form of a hologram. The encoding ...
  70. [70]
    [PDF] Experimental demonstration of coherent photonic neural computing ...
    Nov 1, 2023 · A photonic neural network based on the WDM scheme uses different wavelengths to encode the input signals. That is to say, each photonic synapse ...
  71. [71]
    Parallel optical computing capable of 100-wavelength multiplexing
    Jun 17, 2025 · We propose and demonstrate a parallel optical computing architecture powered by a soliton microcomb source, a broadband Mach–Zehnder interferometer (MZI) mesh.Missing: MIT | Show results with:MIT
  72. [72]
    The Physics of Optical Computing. A New Era After Digital Electronics
    Nov 8, 2023 · Optics can readily support fan-in and fan-out of over 1000 channels, whereas electrical fan-in/out is typically limited to below 10 channels.
  73. [73]
    Programmable optical encryption using thickness-controlled ...
    Mar 26, 2025 · We propose a chiral liquid crystal elastomer (CLCE) designed for stretching-induced multi-wavelength control to enhance photonic encryption functionality.
  74. [74]
    Researchers combine holograms and AI to create uncrackable ...
    Researchers have developed a new optical system that uses holograms to encode information, creating a level of encryption that traditional methods cannot ...
  75. [75]
    Noncoded shadow-casting logic array - Optica Publishing Group
    The use of four sets of individual SLMs is to utilize fully the capability of MOSLM and LCTV to form a negative image by simply rotating the analyzer by 90°.
  76. [76]
    Research Article Optical Computing: A 60-Year Adventure
    Optical computing is a very interesting 60-year old field of research. This paper gives a brief historical review of the life of optical computing from the ...Missing: automata xeroxing
  77. [77]
    All-optical machine learning using diffractive deep neural networks
    Here we introduce a physical mechanism to perform machine learning by demonstrating an all-optical diffractive deep neural network (D 2 NN) architecture.
  78. [78]
    All-optical diffractive neural networks process broadband light
    Dec 2, 2019 · UCLA researchers have designed diffractive networks that can process information using a continuum of wavelengths, expanding this all-optical computation ...
  79. [79]
    Diffractive networks improve optical image classification accuracy
    Jan 11, 2021 · UCLA team reports a leapfrog advance in D 2 NN-based image classification accuracy through ensemble learning.
  80. [80]
    Coherent Ising machines—optical neural networks operating at the ...
    Dec 15, 2017 · In this article, we will introduce the basic concept and the quantum feature of a novel computing system, coherent Ising machines, and describe their ...
  81. [81]
    Quantum Computing for Transport Network Optimization - MDPI
    The coherent Ising machine (CIM), a specialized optical quantum computer using a photonic dissipative architecture, has shown its remarkable computational power ...
  82. [82]
    Performance evaluation of coherent Ising machines against ...
    The coherent Ising machine is expected to find a near-optimal solution in various combinatorial optimization problems, which has been experimentally ...
  83. [83]
    An optical Fourier transform coprocessor with direct phase ... - Nature
    Oct 20, 2017 · We propose and demonstrate a practical method to optically evaluate a complex-to-complex discrete Fourier transform. By implementing the Fourier ...
  84. [84]
  85. [85]
    Photonic processor could enable ultrafast AI computations with ...
    Dec 2, 2024 · Researchers demonstrated a fully integrated photonic processor that can perform all key computations of a deep neural network optically on the chip.<|separator|>
  86. [86]
    [PDF] Intensity-dependent refractive index Optical Kerr effect. - CREOL #ucf
    Optical fiber: Fused silica (SiO2) n2=3.2×10−16 cm2/W. ∆𝜑 = Δ kL ... GW/cm2 for (10µm)2 core, this corresponds to the power of 1.56 kW. - can be ...Missing: threshold | Show results with:threshold
  87. [87]
    Nonlinear Length - RP Photonics
    Jul 13, 2021 · The first part of the equation ( 2 π / λ ) gives you the linear phase change per unit length in vacuum (i.e., for n = 1).
  88. [88]
    Nonlinear optical computing doesn't need nonlinear optics
    Oct 1, 2024 · As adept as optical computing is with linear computations, though, it struggles mightily with nonlinear ones—a small but necessary ingredient ...
  89. [89]
    Electro-Optic Crosstalk in Parallel Silicon Photonic Mach-Zehnder ...
    Jan 4, 2018 · However, the EO crosstalk due to high integration density may degrade the signal transmission performance and eventually limit the possible ...
  90. [90]
    Crosstalk reduction of integrated optical waveguides with ... - Nature
    Mar 11, 2020 · Suppression of the crosstalk between adjacent waveguides is important yet challenging in the development of compact and dense photonic ...
  91. [91]
    Why Optical Computing Still Isn't Mainstream (Yet)? - AZoOptics
    May 19, 2025 · Optical computing uses light for data processing and transmission, offering faster speeds and lower energy consumption than conventional electronic computing.What Is Optical Computing? · The Roadblocks To Widespread... · References And Further...Missing: scholarly | Show results with:scholarly<|control11|><|separator|>
  92. [92]
    Silicon Photonics Manufacturing Ramps Up
    Mar 21, 2024 · The promise of photonics ICs is spurring innovation, but complex processes and a lack of open foundries are keeping it from reaching its full potential.
  93. [93]
    Low-loss fiber-to-chip interface for lithium niobate photonic ...
    So far, the fiber-to-chip loss (commonly >10 dB/facet) has dominated the total insertion losses of typical LN photonic integrated circuits, where on-chip losses ...
  94. [94]
    Cryogenic Systems suitable for Quantum Research - ICEoxford
    High cooling power at 1.0K to allow high efficiency single photon detectors to work alongside optical quantum information processing experiments. · <10mK systems ...
  95. [95]
    Integrated Neuromorphic Photonic Computing for AI Acceleration
    Oct 21, 2025 · Photonic neural networks (PNNs) have achieved influential milestones in AI acceleration, demonstrating single-chip integration of both inference ...
  96. [96]
    Integrated Photonics Research, Silicon and Nanophotonics (IPR)
    This talk will share recent DARPA program investments for increasing the size and complexity of photonic integrated circuits. It will also discuss the ...Missing: PROCEED | Show results with:PROCEED
  97. [97]
    NEUROPULS
    NEUROPULS will develop -for the first time- secure hardware accelerators based on novel neuromorphic architectures and PUF-based security layers.
  98. [98]
    PROgraMmable integrated photonic nEuromorphic and quanTum ...
    Oct 22, 2025 · PROgraMmable integrated photonic nEuromorphic and quanTum networks for High-speed imaging, communications and sEcUrity applicationS
  99. [99]
    2025 IEEE Study Leverages Silicon Photonics for Scalable and ...
    Apr 9, 2025 · Researchers have developed a new hardware platform for AI accelerators capable of handling significant workloads with reduced energy requirement ...Missing: computing | Show results with:computing
  100. [100]
    Bio-inspired optoelectronic devices and systems for energy-efficient ...
    Jul 3, 2025 · This review proposes bio-inspired energy-efficient in-sensor computing utilizing emerging optoelectronic memristors, examining neural network architectures.Missing: PICs | Show results with:PICs
  101. [101]
    2D computational photodetectors enabling multidimensional optical ...
    Jul 23, 2025 · The synaptic vision sensors mimic the plasticity observed in biological synapses, where output current amplification depends on the interval of ...
  102. [102]
    Photonics for sustainable AI | Communications Physics - Nature
    Oct 14, 2025 · Photonic chips can perform fast and energy-efficient General Matrix-Matrix Multiplication (GEMM) operations, which constitute over 90% of the ...
  103. [103]
    FLEX-SCALE: Advancing optical networks for the 6G revolution
    FLEX-SCALE pioneers optical network technologies that will enable unprecedented data speeds while dramatically reducing energy consumption.Missing: exascale ethical
  104. [104]
    From Efficiency to Sustainability: Exploring the Potential of 6G for a ...
    The main findings of this article highlight the effectiveness of these approaches in enhancing energy efficiency, reducing carbon footprint, and optimizing ...Missing: exascale free-
  105. [105]
    PsiQuantum - PostQuantum.com
    In February 2025, DARPA selected PsiQuantum (along with Microsoft) to proceed ... show a convincing small-scale integrated photonic quantum processor working.
  106. [106]
    Lightmatter Raises $400M Series D; Quadruples Valuation to $4.4B ...
    Oct 16, 2024 · Lightmatter, the leader in photonic supercomputing, announced today it has raised a $400 million Series D, valuing the company at $4.4 billion.
  107. [107]
    How Lightmatter Breaks Bandwidth Bottlenecks With Silicon Photonics
    Jan 4, 2024 · Lightmatter's Passage is a silicon photonics interposer designed to support high speed chip-to-chip and node-to-node communications.
  108. [108]
    Lightmatter shows new type of computer chip that could reduce AI ...
    Apr 9, 2025 · Lightmatter revealed on Wednesday it had developed a new type of computer chip that could both speed up artificial intelligence work and use less electricity ...
  109. [109]
    Ayar Labs, with Investments from AMD, Intel Capital, and NVIDIA ...
    Dec 11, 2024 · Ayar Labs, the leader in optical interconnect solutions for large-scale AI workloads, today announced it has secured $155 million in financing.
  110. [110]
    Ayar Labs and Alchip to Scale AI Infrastructure with Co-Packaged ...
    Sep 7, 2025 · “Ayar Labs' co-packaged optics technology unlocks the next era of AI infrastructure by removing the limitations of copper interconnects,” said ...
  111. [111]
    Chip giants again back Ayar Labs for optical interconnects
    Dec 17, 2024 · AMD, Intel, and Nvidia return to join $155M series D funding round intended to revolutionize AI computing infrastructure.
  112. [112]
    PsiQuantum Raises $1 Billion to Build Million-Qubit Scale, Fault ...
    Sep 10, 2025 · PsiQuantum today announced it has raised $1 billion in funding for its Series E round to build the world's first commercially useful, ...
  113. [113]
    PsiQuantum Breaks Ground on America's Largest Quantum ...
    Sep 30, 2025 · PsiQuantum today broke ground on the company's site at the Illinois Quantum and Microelectronics Park (IQMP) in Chicago, where America's ...Missing: commercial | Show results with:commercial
  114. [114]
    PsiQuantum claims silicon photonics breakthrough for quantum ...
    Feb 27, 2025 · PsiQuantum, the Silicon Valley photonic quantum computing startup, is claiming a significant breakthrough with a new high-volume process.Missing: commercial | Show results with:commercial
  115. [115]
    Celestial AI Announces Photonic Fabric™ Adoption by Lead ...
    Mar 24, 2025 · Celestial AI will be at Booth 5739 at OFC 2024, March 25-28, 2024, in the San Diego Convention Center. About Celestial AI. Celestial AI is the ...Missing: processor | Show results with:processor
  116. [116]
    Celestial AI Raises $175M to Commercialize Photonic Fabric
    Celestial AI, creator of the Photonic Fabric optical interconnect technology platform, has raised $175 million in a series C funding round.Missing: co- | Show results with:co-<|separator|>
  117. [117]
    Celestial AI Closes $255 Million Series C1 Funding Round
    Aug 15, 2025 · Celestial AI raised $255 million in Series C1, bringing total funding to $520 million. Funds will accelerate Photonic Fabric deployment and R&D.
  118. [118]
    Frequently Asked Questions: CHIPS Act of 2022 Provisions and ...
    The CHIPS Act of 2022 appropriates $52.7 billion in emergency supplemental appropriations for semiconductor-related programs and activities for FY2023 through ...
  119. [119]
    CHIPS Act funding to be awarded to Coherent, Skywater, and X-Fab
    Dec 17, 2024 · The proposed CHIPS funding would bolster supply resiliency for critical infrastructure markets that were adversely impacted by foundry capacity ...
  120. [120]
    Newsroom 07-22-2025 - Avicena
    Jul 22, 2025 · Avicena focuses on microLED-based interconnects for AI, with a modular LightBundle platform, and has debuted a 1Tbps transceiver and a scalable ...Missing: milestones | Show results with:milestones
  121. [121]
    Highest Power Efficiency and Bandwidth Density - Broadcom CPO
    Benefits. Greater than 3.5x Power Consumption Savings. 40% Lower Optics Cost per Bit. > 1 Tbps/mm Bandwidth Density ; Advantages. Proven Execution · Highest ...Co-Packaged Optics (cpo) · > 1 Tbps/mm Bandwidth... · Ethernet Switch Cpo
  122. [122]
    Photonic Switches Promise to Keep GPUs Fed, Cool - HPCwire
    When it ships in 2026, the Spectrum-X Photonics switch is expected to deliver 200 gigabits per lane of data connectivity on Ethernet networks, ...
  123. [123]
    An integrated large-scale photonic accelerator with ultralow latency
    Apr 9, 2025 · The accelerator is designed to deliver standard linear matrix multiply–accumulate (MAC) functions, enabling computing with high speed up to 1 GHz frequency and ...
  124. [124]
    Photonic processor could streamline 6G wireless signal processing
    Jun 11, 2025 · MIT researchers developed a photonic AI hardware accelerator designed specifically to handle wireless signal processing, reducing latency.Missing: 2023 | Show results with:2023
  125. [125]
    SoftBank Corp. Collaborates with Cisco to Launch "All optical ...
    Oct 23, 2025 · The first phase was completed in the initial area of Osaka Prefecture in September, 2025. This All optical network for metro networks uses ...
  126. [126]
    Self-driving Vehicles and Fourier-optical Computing: A Way Forward?
    We start by considering one of the most challenging applications for high-performance computing: Self-driving vehicles.Missing: 2024 | Show results with:2024
  127. [127]
    Diverging Paths Toward Autonomous Vehicles
    Sep 20, 2024 · SiLC claims that FMCW lidar is “the ideal platform” for safe autonomous driving, noting that it can detect dark objects, distinguish between ...
  128. [128]
    Efficient computation using spatial-photonic Ising machines with low ...
    Mar 3, 2025 · Realizing portfolio optimization in SPIM architectures requires coupling matrix J to be low rank. This is achieved through low-rank factor ...Missing: optics | Show results with:optics
  129. [129]
    Optical processor aids DNA sequencing - Optics.org
    Feb 13, 2018 · The project results are also said to have shown that the optical computing technology holds significant promise for long-read sequence ...<|control11|><|separator|>