Connect with us

Compute & Silicon

Wafer-Scale Chips: What Happens When a Chip Fills the Whole Wafer

Most chips are cut from wafers in many small pieces. Some designs keep the whole wafer as one giant processor. Here is how wafer-scale chips work in general terms.

Published

on

Macro view of a silicon wafer

Normally, a silicon wafer is cut into many separate chips. A few ambitious designs take the opposite approach. They keep a large portion of the wafer intact and treat it as a single, enormous processor. This idea, called wafer-scale integration, has a long history and has found new interest in AI.

The basic idea

A large AI model is usually split across many separate chips, which then have to send data to each other through packages, circuit boards and cables. Each hop costs time and energy.

A wafer-scale design keeps the computing elements on one continuous piece of silicon, linked by on-wafer wiring. Communication stays on the silicon, where it is much faster and more efficient than going off-chip. The result is a processor with an enormous number of cores and a large amount of on-chip memory in a single unit.

Advertisement

An old idea revisited

Wafer-scale integration is not new. Engineers explored it as early as the 1960s and again in the 1980s, but those efforts ran into serious obstacles, especially defects. With a regular chip, a defect ruins one small die. With a wafer-scale chip, a defect could ruin the whole thing.

Modern wafer-scale designs address this with redundancy. They include spare cores and spare connections throughout the wafer, so defective areas can be routed around. The design assumes defects will happen and tolerates them.

The engineering challenges

Building a working wafer-scale processor requires solving several hard problems:

  • Defect tolerance. Spare elements and flexible routing must handle flaws wherever they occur.
  • Crossing the reticle limit. Lithography tools expose only a limited area at once, so designers need ways to connect circuits across the boundaries between exposures.
  • Power delivery. A wafer-scale chip draws a great deal of power, which must be supplied evenly across its surface.
  • Cooling. All that power becomes heat, which must be removed across a large area.
  • Mechanical stress. Silicon and the materials around it expand differently as they heat up, so packaging must handle that movement.

Solving these problems requires custom packaging, cooling and system design, so wafer-scale chips are typically sold as complete systems rather than as components.

Strengths and tradeoffs

The main strength of a wafer-scale approach is the huge amount of fast on-chip memory and communication in one place. For workloads that benefit from keeping data on the silicon, that can be a significant advantage.

Advertisement

The tradeoffs include high system cost, specialized infrastructure and the need for software adapted to the architecture. On-chip memory, while very fast, is still limited compared with the external memory attached to many accelerators, so very large models may need additional memory or multiple systems.

Why it is worth knowing about

Wafer-scale chips illustrate a broader trend in AI hardware: the effort to reduce the cost of moving data. Chiplets, advanced packaging, high-bandwidth memory and wafer-scale designs are all different answers to the same question. How do you bring more computing and memory closer together? Watching how each approach develops is one of the more interesting threads in AI hardware.

How Defect Tolerance Works in Practice

The redundancy idea is easiest to understand with a simple picture. Imagine the wafer as a large grid of identical small cores, like tiles on a floor, with connections running between neighbors in every direction.

  1. Build in spares. The design includes more cores and more connection paths than the system needs to advertise. A small percentage are held in reserve.
  2. Test after manufacturing. Each wafer is tested to find which cores or links did not come out right. The locations differ from wafer to wafer, because defects land randomly.
  3. Map around the flaws. Configuration settings mark the bad elements as unused and connect spare elements in their place, so the grid looks complete to the software.
  4. Present a uniform machine. Programs see a consistent array of working cores, even though the physical layout of each wafer is slightly different underneath.

The key design choice is making each core small. If a defect knocks out one tiny core out of a very large number, losing it barely matters. That is the reverse of the old problem, where a single flaw could doom the entire wafer.

Comparing Approaches to the Same Goal

Wafer-scale chips are one of several ways to bring computing and memory closer together. A quick comparison helps place them:

  • Many separate accelerators: the most common approach. Flexible and widely supported, but data must travel off-chip through packages, boards and cables.
  • Chiplets in an advanced package: several smaller dies joined closely, often with stacked memory. Better communication than separate chips while keeping each die small enough for good yield.
  • Wafer-scale: the largest single slab of connected silicon. The most on-silicon communication, with the most specialized engineering.

Each sits at a different point on the tradeoff between flexibility, cost and communication speed. None is universally best; the right fit depends on the workload and the buyer’s infrastructure.

Common Misunderstandings

“A wafer-scale chip has no defects”

It almost certainly does. The design works because it expects and works around them, not because manufacturing is flawless.

“Bigger silicon always wins”

Size brings communication advantages, but performance on a given job also depends on memory capacity, software and how well the model maps onto the hardware.

“It plugs into a normal server”

Because of its power, cooling and packaging needs, a wafer-scale processor generally comes as part of a purpose-built system rather than as a card that slots into a standard machine.

Questions Readers Ask

Why is crossing the reticle limit such a problem?

A lithography tool prints one rectangular field at a time and repeats it across the wafer. Normally, wiring never needs to cross from one field to the next, because each field becomes its own chip. A wafer-scale design needs reliable connections across those boundaries, which requires special design and manufacturing steps.

What kinds of work suit wafer-scale designs?

Workloads that benefit from keeping data on fast on-chip memory and from very quick communication between cores are natural candidates. Workloads that need memory far beyond what fits on the wafer may need additional resources.

What should I watch for in news about these systems?

Look for details on memory capacity, how multiple systems connect together and how easily existing software runs on them. Those factors shape where the approach fits.

This explainer is general education and does not evaluate any company as an investment. For financial decisions, 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