362 Gbps Over Light: How a Laser Chip Could Reshape Indoor Wireless

A Tiny Chip With an Outsized Signal

The wireless industry has spent decades squeezing more data through radio waves. Wi-Fi 7, the latest standard, pushes theoretical peak speeds toward 46 Gbps. But a research team led by the University of Cambridge just demonstrated something that makes even that number look modest: a chip smaller than a millimeter that transmitted 362.7 gigabits per second over infrared light — using roughly half the energy of comparable Wi-Fi systems.

The result, published in the peer-reviewed journal Advanced Photonics Nexus in March 2026, is among the highest data rates ever achieved by a chip-scale optical wireless transmitter. It doesn't just inch past existing records; it reframes the conversation about how indoor spaces will handle the tidal wave of data demand that AI workloads, 8K streaming, and dense IoT deployments are creating.

The Technology: 25 Lasers on a Pinhead

At the heart of the system is a 5×5 array of vertical-cavity surface-emitting lasers, or VCSELs. These are not exotic laboratory curiosities. VCSELs are the same class of infrared semiconductor laser already embedded in hundreds of millions of smartphones for Face ID, in data center transceivers shuttling petabytes between servers, and in the 3D-sensing modules of autonomous vehicles. What makes them compelling is that they can be manufactured in dense arrays using standard semiconductor fabrication — the same industrial processes that produce the chips in your phone.

The entire 25-laser array in this study fits on a chip smaller than one millimeter, according to the researchers. During testing, 21 of the 25 lasers were operational, each independently transmitting between 13 and 19 Gbps of data over a two-meter free-space link. Each laser is individually addressable, meaning it carries its own independent data stream — a feature that enables true spatial multiplexing, where different users or devices can be served by different beams simultaneously.

To put the per-laser figure in context: a single one of these lasers, at its peak, approaches the throughput of an entire high-end Wi-Fi 7 channel.

Beam Shaping: The Optics That Make It Work

Raw laser power means nothing if the light scatters uselessly across a room. The Cambridge team's key engineering contribution is a precision beam-shaping system that controls exactly where each laser's light goes.

The optical design pairs a custom microlens array — one tiny lens aligned to each VCSEL — with additional lenses that redistribute the 21 individual beams into a structured grid of square illumination spots at the receiving surface. According to the researchers, this arrangement achieved greater than 90 percent beam uniformity across the illuminated area at a distance of two meters, with minimal overlap between adjacent beams.

Why does uniformity matter? In any wireless system, an uneven signal means some spots get excellent service while others suffer. The structured-grid approach ensures that each beam delivers a consistent, predictable signal to a specific zone. This is a fundamentally different architecture from Wi-Fi, where a single radio antenna floods an area with overlapping signals that interfere with each other. By contrast, each laser beam is a clean, directed channel — more like a dedicated fiber-optic line through the air than a broadcast radio signal.

Energy: The Quiet Revolution

The speed figure grabs headlines, but the energy number may matter more in the long run. The system consumed approximately 1.4 nanojoules per bit, which the researchers describe as "roughly half" that of state-of-the-art Wi-Fi technologies under comparable conditions.

This efficiency advantage comes from the physics of VCSELs themselves. Unlike Wi-Fi radios, which require complex signal processing chains to encode data onto radio waves and amplify them through walls and around furniture, VCSELs convert electrical signals directly into modulated light with minimal overhead. The laser sources are, as Optics.org reported, "naturally power efficient" and capable of high-speed operation without the elaborate power amplification stages that dominate Wi-Fi energy budgets.

For a single access point, the energy savings might seem incremental. But scale the math to a modern office building with hundreds of access points, a data center with thousands, or a hospital campus operating around the clock, and that factor-of-two difference in energy per bit becomes a meaningful reduction in electricity consumption and cooling costs. In an era when data center energy use is under intense scrutiny — driven by the explosive growth of AI training and inference workloads — halving the energy cost of internal data transport is not a minor optimization.

Multiuser Reality Check

The 362.7 Gbps aggregate figure represents all 21 lasers firing at maximum capacity toward a single high-bandwidth receiver. Real-world deployment will look different. To explore this, the team ran a separate multiuser demonstration: four simultaneous beams serving four independent receivers delivered a combined approximately 22 Gbps.

That number deserves honest framing. At roughly 5.5 Gbps per user, it significantly exceeds what most current Wi-Fi deployments deliver in practice — real-world Wi-Fi 7 speeds are typically well below the theoretical 46 Gbps maximum — but it's a far cry from the headline aggregate. The gap reflects a basic engineering trade-off: dedicating more lasers to a single user maximizes throughput, while distributing them across users maximizes coverage.

The researchers also acknowledged a significant hardware constraint. As EurekAlert noted, the demonstrated speeds were "limited by the bandwidth of the commercial photodetector" used as the receiver. In other words, the transmitter was outrunning its receiver. According to Interesting Engineering, the team indicated that "speeds could go even higher with faster receivers" — suggesting the transmitter architecture has headroom that current off-the-shelf components cannot yet exploit.

What This Is — and What It Isn't

It is tempting to frame this as "the end of Wi-Fi." The researchers themselves explicitly pushed back on that narrative. As ScienceDaily reported, the technology is "not meant to replace Wi-Fi or cellular networks" but rather to work "alongside them, handling high-capacity data traffic in indoor environments."

This distinction matters. Optical wireless has inherent constraints that radio does not. Light cannot penetrate walls. It requires a clear line of sight between transmitter and receiver (though the beam-shaping grid relaxes the alignment precision needed). It works indoors but not outdoors in direct sunlight or rain. These are not engineering problems to be solved; they are physics.

But for the specific problem of high-density indoor connectivity — an office floor where dozens of employees stream video calls and access cloud applications simultaneously, a hospital ward where real-time imaging data moves between scanners and workstations, a factory floor dense with IoT sensors — optical wireless offers something radio fundamentally cannot: interference-free, directed, high-bandwidth channels that do not compete with each other for shared spectrum.

The potential integration points are practical: ceiling-mounted access points resembling light fixtures, according to Optics.org, that could be deployed alongside conventional Wi-Fi in spaces where demand exceeds what radio can deliver.

The Broader Li-Fi Lineage

This research does not exist in isolation. It builds on more than a decade of work in optical wireless communication, a field most commonly associated with "Li-Fi" — a term coined by Harald Haas, who is among the co-authors of this study.

Haas, a professor at the University of Edinburgh and Director of the Li-Fi Research and Development Center, first demonstrated the concept of using commercial LED light bulbs as broadband wireless transmitters in a TED Global talk in 2011. TIME Magazine recognized the technology as one of the 50 best inventions of that year. In the years since, Haas has published over 300 peer-reviewed papers and holds 31 patents in the field.

Early Li-Fi prototypes used LEDs and achieved data rates in the megabit-per-second range. The leap to VCSELs — semiconductor lasers rather than LEDs — represents a generational shift. VCSELs can be modulated at far higher frequencies than LEDs, and their coherent light can be shaped and directed with much greater precision. The 362.7 Gbps result is, in a sense, the payoff of a 15-year research arc from proof-of-concept LED demonstrations to a chip-scale laser system that rivals the throughput of fiber-optic trunk lines.

The collaboration itself reflects the maturity of the effort. The research team spans the University of Cambridge, University of Manchester, Integrated Compound Semiconductors Ltd, and Compound Semiconductor Centre Ltd — combining academic photonics expertise with the industrial semiconductor fabrication capability needed to move from lab to product.

The Road From Lab Bench to Ceiling Tile

Every laboratory breakthrough faces the same question: what happens when it meets the real world? Several challenges stand between this demonstration and commercial deployment.

Receiver technology is the most immediate bottleneck. The current system used a commercial photodetector that could not keep up with the transmitter's full capacity. Purpose-built receivers — potentially using arrays of high-bandwidth photodiodes matched to the transmitter's beam grid — will be necessary to unlock the system's full potential.

Room-scale coverage presents a different kind of engineering problem. The current demonstration covered a two-meter link. Filling an entire office floor with structured-grid optical coverage will require networks of ceiling-mounted transmitters with handoff protocols, much as cellular networks manage handoffs between towers.

Standards and interoperability remain nascent. IEEE 802.11bb, a standard for light-based wireless communication, was published in 2023, but it was designed around LED-based systems with far lower data rates. VCSEL-based systems operating at hundreds of gigabits per second may require new protocol work.

Cost is unknowable at this stage. VCSELs themselves are cheap to manufacture — the smartphone industry produces billions annually — but the precision beam-shaping optics, high-speed receivers, and room-scale deployment infrastructure add layers of expense that only volume production can drive down.

None of these challenges are fundamental barriers. They are engineering and market problems — the kind that typically yield to sustained investment once the underlying physics is proven.

What to Watch For

This research marks a clear inflection point. A chip-scale optical wireless system has, for the first time, achieved aggregate throughput that rivals short-reach fiber-optic links while consuming dramatically less energy than radio-based alternatives.

The implications extend beyond wireless networking. Data centers — where the energy cost of moving data between racks is a growing concern — could use VCSEL-based free-space optical links as an alternative to the dense bundles of fiber and copper that currently connect servers. Augmented reality systems that demand multi-gigabit, low-latency data streams could benefit from directed optical links that bypass the congestion of shared radio spectrum.

The next milestones to watch: receiver technology catching up to the transmitter, room-scale demonstrations beyond two meters, and the first commercial pilot deployments — likely in enterprise or data center environments where the density and cost constraints favor optical solutions.

The physics is proven. The engineering is underway. The question is no longer whether light can carry the load — it is how quickly the industry can build the infrastructure to let it.

Key Takeaways

• Record throughput: A 25-VCSEL chip smaller than a millimeter achieved 362.7 Gbps over a two-meter infrared link, with 21 lasers active — among the highest chip-scale optical wireless speeds reported.
• Half the energy: The system consumed approximately 1.4 nanojoules per bit, roughly half that of state-of-the-art Wi-Fi, with efficiency gains rooted in VCSEL physics rather than software optimization.
• Complement, not replacement: The researchers position optical wireless as a supplement to Wi-Fi and cellular for high-density indoor environments — not a wholesale replacement, given line-of-sight requirements.
• Receiver is the bottleneck: The transmitter outperformed its commercial photodetector receiver, suggesting significant headroom for higher speeds as receiver technology advances.
• From LED to laser: The work extends Harald Haas's Li-Fi vision from LED-based proofs of concept to VCSEL-based systems delivering fiber-optic-class throughput — a 15-year research trajectory reaching a new milestone.