Find IP Sell IP AI Assistant Chip Talk Chip Videos About Us
Log In

Chip Talk > The Natural Cycle of the Semiconductor World: Explaining the NPU Population Collapse

The Natural Cycle of the Semiconductor World: Explaining the NPU Population Collapse

Published June 13, 2025

Introduction: The Familiar Pattern in Semiconductor IP

For those immersed in the lifecycle of semiconductor IP, the "Boom and Bust" cycle is a well-acknowledged phenomenon. The industry has frequently witnessed a surge in new players and tech innovations followed by a consolidation phase where many companies fade away. The current situation with Neural Processing Units (NPUs) is no different, paralleling past trends within our domain. For more insights on this pattern, explore here.

Recapping Historical Shakeouts

We've seen these patterns before with CPUs and DSPs. Across the late 1990s into the 2000s, a horde of CPU and DSP architectures vied for dominance. However, limited by investment capital and skilled programming talent, the market collapsed into just a few key players. Companies like Tensilica and Arc were eventually acquired by larger entities, and similar dynamics unraveled in the NPU realm.

NPU Market Dynamics

The advent of Artificial Intelligence (AI) sparked the birth of a multitude of NPU-focused companies. From 2018 to the present, an influx of startups aimed to capitalize on embedding AI capabilities in devices. As with previous waves, investments were initially abundant. However, the rapid evolution of AI models, such as the shift toward transformer models and large language models, soon made many first-generation NPUs inadequate.

What Drives 'Resource Scarcity' for NPUs?

When minor players in the NPU market find themselves on rocky ground, it's often due to a shortage of investment and expertise—particularly for compiler development. Without compelling investments secured by demonstrated market traction, startups succumb to a competitive environment. That very scarcity is evident with investors becoming more discerning, pressuring startups for tangible business growth amid tightening capital streams.

Forest or Desolate Isle: What Lies Ahead?

As costs and complexities rise in the newer realms of AI, only a select few of the current 50+ NPU firms might become successful monopolies with potent software and business strategies. The survivors will likely have advanced software toolchains, seamless user programmability for AI models, and abundant venture capital access.

Conclusion: Identifying the Emerging Leaders

For experts in the field, the key to thriving in this era of consolidation lies in identifying the companies showcasing these success markers. One example is Quadric.io, which is anticipated to be a robust contender in the landscape awaiting an NPU collapse reset. By acknowledging historical patterns, professionals can gain confident foresight into the ongoing and future cycles within the semiconductor IP industry.

Get In Touch

Sign up to Silicon Hub to buy and sell semiconductor IP

Sign Up for Silicon Hub

Join the world's most advanced semiconductor IP marketplace!

It's free, and you'll get all the tools you need to discover IP, meet vendors and manage your IP workflow!

Sign up to Silicon Hub to buy and sell semiconductor IP

Welcome to Silicon Hub

Join the world's most advanced AI-powered semiconductor IP marketplace!

It's free, and you'll get all the tools you need to advertise and discover semiconductor IP, keep up-to-date with the latest semiconductor news and more!

Plus we'll send you our free weekly report on the semiconductor industry and the latest IP launches!

Switch to a Silicon Hub buyer account to buy semiconductor IP

Switch to a Buyer Account

To evaluate IP you need to be logged into a buyer profile. Select a profile below, or create a new buyer profile for your company.

Add new company

Switch to a Silicon Hub buyer account to buy semiconductor IP

Create a Buyer Account

To evaluate IP you need to be logged into a buyer profile. It's free to create a buyer profile for your company.

Chatting with Volt