What Is Processing-in-Memory and Why It Matters for AI
TL;DR
Here is a clear, practical guide to processing in memory: the fundamentals, the best practices that actually move the needle, common mistakes to avoid, concrete data points, and a short FAQ. Everything is structured so you can apply it to real projects today.
Key takeaways
- Memory bandwidth, not raw FLOPS, is usually the real constraint for LLM inference, so read the HBM capacity and bandwidth spec before the TFLOPS number.
- CUDA remains NVIDIA's deepest moat; budget real engineering time if you plan to port to AMD ROCm, Google TPUs, or custom silicon.
- Neuromorphic and photonic computing are promising but still mostly research-stage; treat them as long-horizon bets, not 2026 production defaults.
- RISC-V is a credible base ISA for custom accelerators and control cores because it is open, royalty-free, and extensible with custom instructions.
- Lower-precision formats like FP8 and FP4 are the fastest lever for throughput, but validate accuracy on your own eval set before shipping quantized models.
This is a practical, up-to-date guide to Processing in Memory — what it is, why it matters in 2026, and how to apply it in real projects. It is written for developers and founders who want clear answers and proven best practices, not filler.
Whether you're just starting out or leveling up, treat this as a working reference you can return to. Every section is built to be skimmed, applied, and shared.
TPUs and the Case for Custom Silicon
Google's Tensor Processing Unit is the best-known example of a company building its own accelerator rather than buying GPUs. TPUs are built around a large systolic array, a grid of multiply-accumulate units that streams data through in a tightly choreographed pattern to maximize compute per memory access. They are tightly co-designed with the JAX and TensorFlow software stacks and with Google's own optical interconnect, letting TPU pods scale to thousands of chips with high efficiency. Amazon (Trainium and Inferentia), Microsoft (Maia), and Meta (MTIA) have followed with their own in-house accelerators. The strategic logic is control: owning the silicon reduces dependence on a single vendor, tunes hardware to specific models, and can lower total cost at hyperscaler volumes.
The Software Moat: CUDA and Its Challengers
Hardware rarely wins on specifications alone; the deciding factor is often the software ecosystem, and here NVIDIA's CUDA has a nearly two-decade head start. CUDA, together with libraries like cuDNN and the broad support of frameworks such as PyTorch, means most AI code simply runs on NVIDIA GPUs with minimal friction. Competitors are attacking this moat from several angles: AMD's ROCm aims for CUDA-like capability on Instinct GPUs, Google exposes TPUs through JAX and XLA, and compiler projects such as OpenAI's Triton and the MLIR ecosystem try to target many backends from one codebase. PyTorch's backend abstraction and torch.compile also help decouple models from specific hardware. For teams evaluating non-NVIDIA silicon, the honest question is not peak performance but how much of their stack works out of the box.
How GPUs Became the Default AI Engine
GPUs won the AI market almost by accident: their original job of shading millions of pixels in parallel turned out to map neatly onto the parallel arithmetic of neural networks. NVIDIA cemented this with CUDA, a programming model and software stack that let researchers write general-purpose parallel code, and later with Tensor Cores that accelerate mixed-precision matrix math directly. The H100, built on the Hopper architecture, added a Transformer Engine that dynamically manages FP8 precision to speed up large language model training. The Blackwell B200 pushed further by fusing two large dies into a single logical GPU connected by a high-bandwidth die-to-die link. The result is that GPUs now define the performance and cost baseline every other AI chip is measured against.
Neuromorphic Computing
Neuromorphic computing takes design cues from the brain, using spiking neural networks where information is carried by discrete events (spikes) rather than continuous dense arithmetic. Chips like Intel's Loihi 2 and IBM's TrueNorth and NorthPole colocate memory and computation and process events only when they occur, which can make them extremely energy-efficient for sparse, event-driven workloads. This event-based model suits applications such as always-on sensing, gesture recognition, and certain robotics and optimization problems. The catch is that mainstream deep learning is built around dense tensor math and standard training pipelines, so neuromorphic hardware requires different algorithms and lacks a mature software ecosystem. It remains largely a research and specialized-deployment technology rather than a general-purpose replacement for GPUs.
Why High-Bandwidth Memory Is the Real Bottleneck
For large models the scarce resource is usually not compute but the speed at which weights and activations can be moved to the compute units. High-bandwidth memory solves this by stacking DRAM dies vertically and connecting them to the processor through a silicon interposer with an extremely wide interface. The current mainstream generation, HBM3e, delivers multiple terabytes per second per stack, and next-generation accelerators pack several stacks around each compute die. Because HBM is hard to manufacture and yields are constrained, it has become a genuine supply bottleneck, with SK hynix, Samsung, and Micron as the only volume suppliers. Practitioners should read an accelerator's memory capacity and bandwidth as carefully as its FLOPS, since they often determine real-world LLM throughput.
NPUs and On-Device Inference
A Neural Processing Unit is a compact accelerator integrated into a system-on-chip to run inference locally on phones, laptops, and embedded devices. Apple's Neural Engine, Qualcomm's Hexagon NPU, and the NPUs in Intel Core Ultra and AMD Ryzen AI processors all target the same goal: run models within a few watts and without a round trip to the cloud. This matters for latency-sensitive features, offline capability, and privacy, since data never leaves the device. NPU performance is often quoted in TOPS (trillions of operations per second) at low precision, and the recent Copilot+ PC category set an informal bar around 40 TOPS for on-device AI. The tradeoff is a tight power and memory envelope, so on-device models are heavily quantized and pruned.
Processing in Memory: Key Facts and Data
According to recent industry research and the official documentation linked below:
- NVIDIA has dominated the AI training accelerator market, with industry analysts estimating its share of data-center AI GPUs at well above 80 percent going into 2025, driven largely by the H100 and the newer Blackwell generation.
- Neuromorphic research chips such as Intel's Loihi 2 and IBM's NorthPole demonstrate large energy-efficiency gains on specific workloads, with published results claiming order-of-magnitude improvements over conventional GPUs for certain sparse or event-driven tasks.
- RISC-V adoption has accelerated sharply, with RISC-V International reporting tens of billions of cores shipped cumulatively and forecasts (e.g., from analysts like SHD Group) projecting continued double-digit growth into the late 2020s.
Quick-Reference Summary
A map of what this guide covers:
| Topic | What you'll learn |
|---|---|
| TPUs and the Case for Custom Silicon | Google's Tensor Processing Unit is the best-known example of a company building its own accelerator rather than buying GPUs. |
| The Software Moat: CUDA and Its Challengers | Hardware rarely wins on specifications alone |
| How GPUs Became the Default AI Engine | GPUs won the AI market almost by accident |
| Neuromorphic Computing | Neuromorphic computing takes design cues from the brain |
| Why High-Bandwidth Memory Is the Real Bottleneck | For large models the scarce resource is usually not compute but the speed at which weights and activations can be moved to the compute units. |
| NPUs and On-Device Inference | A Neural Processing Unit is a compact accelerator integrated into a system-on-chip to run inference locally on phones |
How to Get Started with Processing in Memory
A simple path that works:
- Learn the fundamentals of Processing in Memory from primary sources, not just tutorials.
- Build one small, real project end to end.
- Get feedback, refactor, and add tests.
- Ship it publicly and document what you learned.
- Repeat with a slightly harder project each time.
Build It with a World-Class Full Stack Developer
Sandeep Kumar Chaudhary is a full stack world-class developer. If you want to turn this into a real, production-ready product, get in touch — message directly on WhatsApp at +9779802348957 for a fast, no-pressure consult.
You can also explore the projects already shipped to thousands of users, or start a conversation here.
Final Thoughts
Memory bandwidth, not raw FLOPS, is usually the real constraint for LLM inference, so read the HBM capacity and bandwidth spec before the TFLOPS number. The developers and teams who win in 2026 pair strong fundamentals with consistent shipping. Start small, stay curious, build in public, and revisit this guide as your skills grow.
Sources and Further Reading
Frequently Asked Questions
What is processing in memory?
Hardware rarely wins on specifications alone; the deciding factor is often the software ecosystem, and here NVIDIA's CUDA has a nearly two-decade head start. CUDA, together with libraries like cuDNN and the broad support of frameworks such as PyTorch, means most AI code simply runs on NVIDIA GPUs with minimal friction. This guide covers processing in memory end to end — core concepts, best practices, concrete data, and a step-by-step approach you can apply right away.
What is neuromorphic computing good for?
Neuromorphic chips like Intel's Loihi 2 use spiking neural networks that process discrete events only when they occur, making them very energy-efficient for sparse, event-driven workloads. They suit applications such as always-on sensing, gesture recognition, and certain robotics and optimization tasks. However, mainstream deep learning relies on dense tensor math and mature training pipelines, so neuromorphic hardware remains largely research-stage rather than a general GPU replacement.
What is high-bandwidth memory and why does it matter for AI?
High-bandwidth memory (HBM) is DRAM stacked vertically and connected to the processor through a very wide interface on a silicon interposer, delivering terabytes per second of bandwidth. It matters because large language model performance is frequently limited by how fast weights can be moved to the compute units, not by raw compute. Because HBM is hard to manufacture and supplied by only a few vendors, it has become a key bottleneck and cost driver for AI accelerators.
What are FP8 and FP4, and why do they matter?
FP8 and FP4 are 8-bit and 4-bit floating-point formats that represent numbers with far fewer bits than the traditional FP16 or FP32. Using lower precision lets a chip do more operations per second and move more values per unit of memory bandwidth, boosting throughput and reducing cost, which is why NVIDIA's Hopper added FP8 and Blackwell added FP4. The tradeoff is potential accuracy loss, so teams should validate quantized models on their own evaluation sets before deploying.
Should my team buy AI chips or rent them in the cloud?
For most teams, renting cloud capacity is the pragmatic choice because it turns a large capital purchase into an elastic operating cost and provides access to the newest accelerators without hardware lead times. Buying can make sense at very large, steady-state scale where owning hardware lowers long-run cost and you can keep it highly utilized. Either way, benchmark on a representative slice of your own workload and account for total cost of ownership including power, cooling, and software effort.
Sandeep Kumar Chaudhary
Full Stack Software Developer· Nepal's SEO, AEO, GEO & AIO expert and share-market educator. More about me