Connect with us

Compute & Silicon

Silicon Photonics and Optical Links: Moving AI Data With Light

As AI clusters grow, copper wires struggle to keep up. Here is how optical connections and silicon photonics work, and why they are getting so much attention.

Published

on

Abstract bokeh lights with fiber optics

AI clusters move staggering amounts of data between chips, servers and racks. Most of that traffic travels as electrical signals over copper. As speeds rise and clusters grow, the industry is increasingly turning to light instead. Silicon photonics, the practice of building optical components with chipmaking techniques, sits at the center of that shift.

Why copper runs out of room

Copper is cheap, reliable and works well over short distances. But as data rates climb, electrical signals lose strength faster as they travel. Pushing them further takes more power and more complex circuitry to clean up the signal. At the highest speeds, the practical reach of copper shrinks to a few meters or less.

In an AI cluster, where thousands of accelerators need fast connections, that limitation shapes how tightly equipment must be packed and how much energy goes into moving data rather than computing.

Advertisement

How optical links work

An optical link converts an electrical signal into light, sends it through a glass fiber and converts it back at the other end. Light in fiber can travel long distances with little loss and can carry enormous amounts of data, especially when multiple wavelengths share the same fiber.

In data centers today, the conversion usually happens in pluggable optical transceivers, small modules that plug into the front of network switches and servers. These have been a workhorse of data center networking for years.

What silicon photonics adds

Silicon photonics builds optical components such as waveguides, modulators and detectors on silicon, using manufacturing methods similar to those used for regular chips. That brings several potential advantages:

  • Components can be made in large volumes using established fabs.
  • Optical and electronic circuits can be placed very close together.
  • Many optical channels can fit in a small area.

Silicon itself does not emit light efficiently, so lasers are typically made from other materials and attached or connected separately. Integrating them reliably is one of the field’s engineering challenges.

Bringing optics closer to the chip

A major direction of research and development is co-packaged optics. Instead of converting signals at the edge of a switch or server, the optical components are placed in the same package as the switch chip or processor. Shorter electrical paths mean less energy lost before the signal becomes light.

Advertisement

Proponents see this as a path to linking more accelerators together with higher bandwidth and lower power. Challenges include heat management, serviceability, since a failed optical part might mean replacing a costly package, and building a supply chain for new kinds of components.

Why it matters for the build-out

Networking is already a large part of what it takes to build an AI cluster, and its share of power use grows as systems scale. Optical technologies offer a way to keep scaling communication without letting energy and physical constraints take over.

For readers, photonics is worth watching because it connects several themes of AI infrastructure at once: the memory and communication bottlenecks of large models, the power limits of data centers and the growing role of packaging in chip progress. Adoption tends to happen gradually, starting at longer distances and moving steadily closer to the chip.

A Simple Way to Picture the Tradeoffs

Choosing between copper and light in a data center is less about which is better and more about which fits each distance. A useful mental map runs from shortest to longest:

  • Inside a package: chips sitting side by side communicate over very short electrical paths. Copper and other conductors dominate here.
  • Within a server or rack: copper cables and circuit board traces often still make sense for short runs, especially where cost and reliability matter most.
  • Between racks: as distances grow to many meters, optical links become the usual choice at high speeds.
  • Across a building or campus: optics are the norm, since copper cannot carry high data rates that far efficiently.

The trend the original article describes is that the boundary between “copper territory” and “optical territory” keeps moving inward as speeds climb. Silicon photonics and co-packaged optics are efforts to make light practical at shorter and shorter distances.

The Building Blocks, in Plain English

Waveguides

These are tiny channels that guide light across a chip, much as a fiber guides it over long distances. They are the optical equivalent of wires.

Modulators

A modulator encodes data onto a beam of light by changing its properties, such as brightness or phase, billions of times per second. It turns electrical ones and zeros into an optical signal.

Detectors

At the receiving end, a photodetector converts the incoming light back into an electrical signal the chip can use.

Lasers

The light source itself. Because silicon is a poor light emitter, lasers are typically made from other semiconductor materials and combined with the silicon parts, either inside the package or in a separate module that feeds light in through fiber.

Common Misunderstandings

  • “Light computes faster.” In these systems, light carries data; the computing still happens in electronic chips. The benefit is in moving information, not in doing math.
  • “Optics will replace copper everywhere.” Copper remains cheaper and simpler for many short links. The two are likely to coexist for a long time.
  • “Co-packaged optics is a finished product category.” It is an active area of development, and approaches differ. Adoption depends on solving reliability, service and manufacturing questions at scale.

What to Watch as the Technology Develops

For readers following optical networking, a few questions help put announcements in context:

  • At what distance is the new optical approach being used: between racks, inside a rack or inside a package?
  • Is the claimed improvement in bandwidth, power per bit or density?
  • How is the laser supplied, and can a failed part be replaced without discarding an expensive chip?
  • Is the product shipping in volume, sampling to customers or still a demonstration?

Questions Readers Ask

Why does power per bit matter so much?

In a large cluster, the energy spent moving each bit adds up across enormous traffic volumes. Small savings per bit can translate into meaningful savings at the scale of a data center.

Is silicon photonics new?

The field has been researched for many years, and silicon photonic components are already used in some data center transceivers. What is newer is the push to bring them right next to processors and switch chips.

This explainer is general education and not investment advice. For financial decisions involving networking or photonics companies, consult a qualified financial professional.

Latest stories

View all

Compute & Silicon

Why a Cold Cache Changes the Meaning of a Fast Response Test

A service that has recently handled a request may respond differently from one meeting it for the first time. Data, compiled work or model state can already be available, so a warm test and a cold test answer different questions.

Published

on

A rack of servers in a server room

Photo: Kevin Ache / Unsplash

A service that has recently handled a request may respond differently from one meeting it for the first time. Data, compiled work or model state can already be available, so a warm test and a cold test answer different questions.

The operating distinction

Caches keep selected information close to where it will be used. Reusing that information can reduce retrieval or preparation work. A cold start may need to rebuild it, load a model or fetch data from a slower location. Neither condition is inherently the only valid test; the expected workload determines the mix.

Advertisement

Preparation can also occur outside the part of the system being timed. A demonstration may load information before starting its stopwatch, while a user experiences that loading as part of the wait. That does not invalidate the demonstration, but it changes the boundary of the claim.

A practical example

Imagine comparing two demonstrations of the same service. One repeats a familiar query while the other begins after a restart. Their response times may differ without any change in the underlying hardware. A fair comparison describes what was already loaded and what preparation was included.

Questions worth asking

A comparison should include preparation that matters to the intended user experience. Keeping warm and cold observations separate makes both useful without presenting the faster result as representative of every possible request.

  • Was the test performed after restart or after repeated requests?
  • Which preparation steps were included in the measured response?
  • How often does the real workload encounter the cold condition?

What the evidence supports

A single fast example cannot establish the experience across restarts, uncommon requests and traffic changes. Conversely, an unusually slow first request should not automatically represent steady operation. Readers need the operating state, the sampling method and the service objective before drawing a conclusion.

A useful comparison can present both first-use and repeated-use observations under clearly described conditions. It should identify whether the test reuses the same request or exercises a broader set of inputs. The reader can then decide which observation resembles the intended application. The aim is an honest description of service behavior, not a contest that rewards whichever test omits the most preparation work.

Advertisement

Treat cache state as part of the test description, alongside the machine and the application.

Quick check

How much did you catch? Tap a question to see the answer.

1. Why do warm and cold tests give different results?

Answer: Data, compiled work or model state may already be loaded in a warm test.

2. Is a cold test the only valid one?

Answer: No. The expected workload decides the right mix.

3. What should a fair comparison describe?

Answer: What was already loaded and what preparation was included in the timing.

Where Warm State Lives in a Modern Service

The word cache can sound like a single component, but a typical service holds several layers of prepared state at once. Each layer can be warm or cold independently, which is one reason response tests can be difficult to compare.

  • Model weights in memory. A model already loaded onto an accelerator is ready to run. A model that must be read from storage first adds loading time.
  • Compiled or optimized code. Some systems prepare optimized execution paths the first time a particular shape of work appears and reuse them afterward.
  • Request-level reuse. Inference systems may reuse computation for repeated prompt prefixes or recently seen inputs.
  • Application data caches. Frequently retrieved documents, records or results may be held in faster memory near the service.
  • Connection and session state. Established network connections, authentication tokens and open sessions avoid setup work for subsequent requests.
  • Storage and operating system caches. Files read recently may already sit in memory, making later reads faster.

Designing a Fairer Response Test

Anyone evaluating a claim about response speed, or designing a test of their own, can work through a short sequence.

  1. Describe the starting state. State whether the test began after a restart, after a period of inactivity or after a run of similar requests.
  2. Define the timing boundary. Record exactly when the clock starts and stops, and which preparation steps fall inside or outside that window.
  3. Vary the inputs. Use a set of requests that resembles the real workload rather than repeating one query.
  4. Report a distribution. Present typical and slower responses, not only the fastest or the average.
  5. Estimate the real cold rate. Consider how often actual users will meet a cold condition, based on traffic patterns, scaling behavior and restarts.
  6. Keep the results labeled. Publish warm and cold observations side by side with their conditions.

Common mistakes

A frequent mistake is running a test many times and reporting the best result, which almost always reflects fully warm conditions. Another is forgetting that automatic scaling can introduce cold starts whenever new capacity is added during a traffic increase. A third is comparing two systems where one test included model loading and the other did not. A fourth is assuming that a cold start happens only after a deliberate restart, when cache eviction under memory pressure can produce similar effects.

What This Means for Buyers and Readers

For a business choosing an inference service or reading a vendor’s performance claim, the cache question is practical. A service that is consistently fast for repeated requests may still feel slow to users who arrive with varied or uncommon requests, or who arrive just after a scaling event. Asking how the provider handles cold conditions, and how often they occur in practice, gives a more realistic sense of what users will experience.

Fast is a description of a particular state. The useful question is how often the real workload is in that state.

Questions Readers Ask

Is a warm test misleading?

Not if it is labeled. Warm results describe steady operation for repeated work, which is useful. They become misleading only when presented as the whole experience.

Can cold starts be eliminated?

They can be reduced by keeping capacity loaded and preparing state in advance, but this usually carries a cost in reserved resources.

Why does the first request after a restart take longer?

The service may need to load models, prepare execution paths, open connections and fill caches before it can respond.

What should a buyer ask a provider?

Ask what conditions their published response figures assume and how the service behaves after restarts, scaling events and long idle periods.

BOSSTOX is coming to iPhone

AI infrastructure explainers and news on iPhone, iPad and Apple Watch.

See BOSSTOX

Continue Reading

Compute & Silicon

An Idle Accelerator Can Still Be Waiting on Useful Work

A low utilization reading does not identify a single problem. An accelerator can be waiting for data, another machine, a scheduling decision or a request from a user. Understanding the waiting state matters more than assigning blame to the chip.

Published

on

Close-up of a circuit board

Photo: Anne Nygård / Unsplash

A low utilization reading does not identify a single problem. An accelerator can be waiting for data, another machine, a scheduling decision or a request from a user. Understanding the waiting state matters more than assigning blame to the chip.

The operating distinction

Useful work passes through several stages before an accelerator starts arithmetic. Data must be available, memory must have room and dependent tasks must finish. A scheduler may also hold capacity for a workload with a deadline. That reserved headroom can look similar to inefficiency on a broad average.

Advertisement

Different kinds of waiting call for different responses. Adding processors will not necessarily repair a slow input pipeline, and changing scheduling can affect one customer's response while improving another's throughput. The resource graph and the service objective need to be read together.

A practical example

Imagine an inference service that keeps spare resources available for sudden requests. Filling every available slot with background work could increase the average utilization while making urgent requests slower. A training job with predictable input and long execution periods would present a different operating problem.

Questions worth asking

Reserved headroom, missing input and inefficient execution are different explanations for unused capacity. The next operational question should follow the observed cause rather than assume that every idle interval represents the same problem.

  • What resource or dependency is the accelerator waiting for?
  • Does the measurement include reserved capacity and maintenance periods?
  • Did higher utilization improve completed work or merely lengthen queues?

What the evidence supports

Utilization needs a definition and a time window. A device-level percentage alone cannot show whether a service met its response target or whether a job completed on schedule. Compare the waiting causes and user outcome before comparing the headline averages.

A useful account identifies the interval when the observation was made and describes the mix of jobs present. Readers can then distinguish an idle system with no demand from a busy service deliberately retaining headroom. Both can show unused device capacity, but they tell different stories about the amount of work waiting and whether customers are receiving the experience the service was designed to provide.

Advertisement

The useful target is productive work within the service requirement, not a percentage detached from its purpose.

Quick check

How much did you catch? Tap a question to see the answer.

1. Does low utilization point to a single problem?

Answer: No. An accelerator may be waiting on data, another machine, scheduling or users.

2. Why can reserved headroom look inefficient?

Answer: It can resemble idle capacity on a broad average.

3. Will adding processors fix a slow input pipeline?

Answer: Not necessarily.

Sorting Idle Time Into Categories

A single utilization line becomes far more useful once the unused intervals are split into a few named categories. The categories do not need to be elaborate. They only need to separate explanations that call for different responses, so that a review meeting can discuss causes instead of a number.

  • No demand. Nothing was waiting. The service was available and nobody needed it during that interval.
  • Deliberate headroom. Capacity was held back on purpose so urgent or latency-sensitive requests could start quickly.
  • Input starvation. Work existed, but data, model weights or request batches were not arriving fast enough to keep the device busy.
  • Coordination waits. The device was ready but waiting on another machine, a synchronization step or a dependent task.
  • Scheduling gaps. Work was queued, yet placement, priority or fragmentation kept it from landing on available hardware.
  • Planned unavailability. Maintenance, testing, updates or recovery activity removed the device from service.

Even a rough split changes the conversation. If most idle time belongs to deliberate headroom, the question becomes whether the reserve is sized correctly for the response target. If most belongs to input starvation, the attention moves to storage, networking or preprocessing. Neither answer requires buying or removing accelerators.

A Step-by-Step Review of Unused Capacity

An operator or analyst trying to understand a low reading can work through a short sequence before drawing conclusions.

  1. Fix the time window. State whether the figure covers an hour, a day, a month or a reporting quarter. Overnight lulls and weekend demand can dominate a long average.
  2. State what counts as busy. Some measurements record whether any work is scheduled on the device, while others try to estimate how much of its arithmetic capability is in use. The two can tell very different stories.
  3. Separate the workload mix. Training, batch processing and interactive inference have different shapes. Averaging them together hides the behavior of each.
  4. Place the waits on a timeline. Line up idle intervals against data transfers, checkpoint writes, job starts and request bursts. Patterns often appear where they would never show in a single percentage.
  5. Check the outcome. Ask whether jobs finished on schedule and whether response targets were met during the same window.
  6. Choose the response that matches the cause. Only then decide whether the answer is more capacity, better scheduling, a faster input path or no change at all.

Common mistakes

The most frequent error is treating utilization as a goal in itself. A team rewarded for a higher figure can fill reserved headroom with low-priority work, which raises the average while making urgent requests wait. Another mistake is comparing utilization across facilities or providers without confirming that the definitions match. A third is ignoring the cost of reaching a higher number, such as longer queues or more frequent preemption of jobs that then have to restart.

What This Means for Readers Following the Industry

Utilization figures appear in earnings discussions, vendor presentations and infrastructure announcements. For a reader trying to understand the business, the figure is a starting point for questions rather than a verdict. A high number may reflect strong demand, or it may reflect a fleet with too little headroom for spikes. A low number may signal weak demand, or a service that keeps capacity in reserve because its customers expect fast responses.

The more useful question is usually what the figure was designed to show and what it leaves out.

This is general education about how infrastructure metrics are read, not a view on any company or security. Anyone weighing a financial decision should consult a qualified financial professional about their own situation.

Questions Readers Ask

Is a fully utilized accelerator always the efficient outcome?

Not necessarily. A device kept constantly busy may leave nothing available for sudden requests. Whether that is efficient depends on the service the device is meant to support.

Can idle time be measured precisely?

It can be measured, but the result depends on the definition, the sampling interval and the tools used. Short waits can disappear inside coarse samples.

Why would a provider intentionally leave capacity unused?

To meet response commitments, absorb failures, allow maintenance or accept new work without delay. Each of those purposes has value that an average cannot display.

What is the single best follow-up question?

Ask what the accelerator was waiting for during the idle intervals. The answer points toward the next useful decision.

BOSSTOX is coming to iPhone

AI infrastructure explainers and news on iPhone, iPad and Apple Watch.

See BOSSTOX

Continue Reading

Compute & Silicon

Custom AI ASICs vs. GPUs

Custom AI ASICs can deliver strong efficiency on a narrow workload, while GPUs remain valuable because their software ecosystem and flexibility absorb uncertainty.

Published

on

Close-up of a graphics card heatsink

The custom-ASIC-versus-GPU debate is often framed as a contest between a cheaper chip and a more expensive one. That misses the real decision. A GPU is a general-purpose accelerator with a large software ecosystem. An application-specific integrated circuit is a narrower machine designed around a workload, a power target, or a system architecture.

The winner depends on whether the workload is stable enough to justify specialization. Training, inference, recommendation, search, and video pipelines do not stress hardware in exactly the same way. A model that is efficient today can change after a new architecture, quantization method, context length, or customer demand appears.

What the GPU buys

A GPU’s first advantage is optionality. It can run different model families, numerical formats, kernels, and frameworks. It can be reassigned when demand shifts. That matters for a cloud provider serving many customers and for a company whose AI roadmap is still moving.

Advertisement

The second advantage is software. Mature compiler tools, libraries, kernels, orchestration systems, and developer familiarity reduce the risk between purchasing hardware and putting it into production. A GPU’s list price is not the whole cost; neither is an ASIC’s lower unit cost if engineers spend months rewriting software.

The third advantage is scale. GPUs can be linked into standard or proprietary fabrics, with established server, cooling, and support ecosystems. A customer can often rent them before it owns them, which converts a large capital decision into a more flexible operating expense.

The GPU’s weaknesses are also clear. The chip may carry circuitry for workloads the customer does not use. Its memory and network design may be optimized for a broad market rather than one application. High demand can create shortages, and power density can push a facility’s cooling and electrical systems beyond their original design.

What the ASIC buys

A custom ASIC can remove general-purpose features and use the saved area or power for the specific operations that matter. It may deliver more performance per watt, better latency, or lower cost per inference when the workload is predictable.

An ASIC can also be part of a broader system strategy. Google and other hyperscalers have built custom accelerators because owning the silicon roadmap can reduce dependence on a merchant vendor and let the company tune hardware to its software stack. Intel described an expanded 2026 collaboration with Google around custom ASIC-based infrastructure processing units that offload networking, storage, and security functions from host CPUs. That is a reminder that “custom AI silicon” is not limited to a chip that performs matrix multiplication.

Advertisement

The costs appear before production

ASIC design requires architecture, verification, physical design, software, packaging, testing, and manufacturing commitments. A company must pay those costs before it knows how many units it will ship.

The minimum efficient volume can be high. If a custom chip saves $1,000 per accelerator but costs $300 million to develop and qualify, the business needs more than 300,000 units before ignoring financing, support, and inventory risk. If the workload changes before that volume is reached, the savings may disappear.

Advanced packaging and high-bandwidth memory can add another constraint. Micron reported in March 2025 that its HBM output for calendar 2025 was sold out and that it was already discussing 2026 demand with customers. A custom ASIC that cannot secure memory, substrates, and packaging is not a complete product.

A practical comparison must use system metrics

Compare the two options at the system level. Measure useful work per rack, tokens per second per watt, cost per million tokens, time to train, memory capacity, software migration effort, and failure-recovery behavior.

Do not compare a peak chip number from a data sheet with a measured production result from another platform. Ask whether the benchmark uses the same model, batch size, sequence length, precision, compiler version, and utilization target.

A GPU may win on time to market even if an ASIC wins on energy. The value of six months of earlier service can be material for a cloud provider. Conversely, an inference fleet with a stable model and high utilization may benefit from a specialized accelerator.

The market is moving toward heterogeneous systems

The likely outcome is not one winner. General-purpose GPUs will remain useful for training, new models, and workloads with uncertain demand. Custom ASICs will take more of the predictable inference, networking, storage, and recommendation work. CPUs, DPUs, and specialized accelerators will handle other parts of the system.

Intel’s Gaudi 3 product materials emphasize standard Ethernet, open software, and a migration path for customers moving from GPU-based systems. That positioning shows the market’s pressure point: customers want acceleration without locking every layer of the stack to one vendor.

The decision rule is simple. Use GPUs when flexibility, software, and deployment speed are worth paying for. Use an ASIC when the workload is stable, volume is large, memory and packaging are secured, and the organization can support a separate software and operations path.

The best analysis treats silicon as a fleet decision, not a chip-shopping exercise. The question is which architecture produces reliable, adaptable work at the lowest total cost over the period in which the workload is likely to remain stable.

There is also a strategic lock-in cost. A GPU fleet may be portable across cloud providers and model frameworks, while a custom accelerator can require a dedicated compiler, kernel library, scheduler, and support team. That lock-in may be acceptable when demand is large and stable. It is dangerous when the business case assumes a model architecture will remain unchanged for five years. The best ASIC programs preserve a fallback path: compatible APIs, exportable model formats, or enough GPU capacity to absorb new workloads while the specialized stack catches up.

Sources

  1. Micron, “Fiscal Q2 2025 Investor Materials,” March 20, 2025: investors.micron.com
  2. TSMC, “2025 Annual Report,” 2026: investor.tsmc.com
  3. Intel, “Intel Gaudi AI Accelerator Products,” accessed October 2026: intel.com
  4. Intel, “Intel and Google Deepen Collaboration to Advance AI Infrastructure,” April 9, 2026: newsroom.intel.com
  5. IEA, “Energy and AI,” April 10, 2025: iea.org

BOSSTOX Research articles are drafted with AI assistance and reviewed by our editors before publication. See our editorial standards.

Continue Reading

Trending