By Eduardo Rodrigues and Jack Fox-Male  Reading time: 10 minutes

Disruption is a word we hear a lot in today’s world.

In the early 1900s, we heard it a little bit less. But even back then, Joseph Schumpeter loved talking about disruption.

The 20th-century economist’s idea of “creative destruction” has transformed how we understand industry disruption cycles today.

And what we do understand is that today these disruption cycles are compressing dramatically, forcing corporations to adapt at unprecedented speed. If they don’t, they fall behind.

As we move into an AI-dominated world, large corporations are facing overwhelming challenges in energy efficiency, compute power, and even international sovereignty – each of which underpins operations at some level.

These are inherently complex problems. Deep tech problems, you could say.

And to deal with deep tech problems, you need deep tech solutions, along with a strong ecosystem of corporates, startups, entrepreneurs, investors, and founders, to accelerate change.

So, what are the most promising chip technologies that will power the AI world of tomorrow, and how can European corporations capitalise on this to thrive in their industry?

Let’s take a closer look.

Neuromorphic chips are leading the way in the AI transformation

Running advanced AI systems on today’s computers consumes monumental amounts of energy. Data centers already consume 1 to 2% of global electricity, and AI workloads are skyrocketing this demand.

A significant share of this is wasted in the constant movement of data between memory and processors. Therefore, efficiency is key to the future of the semiconductor industry. This is where innovative AI chips are redefining the value chain.

Unlike traditional chips, which power everyday devices, AI chips are designed with optimised architecture to allow features like real-time pattern recognition, sensory data processing, and predictive analytics to perform more energy efficiently.

Early innovations in this space included in-memory computing, which reduces energy loss by moving computation closer to memory.

The next frontier, neuromorphic computing, goes one step further: it creates architectures inspired by the human brain, where calculations occur inside memory in a brain-like, energy-efficient way.

Tackling data processing: Reducing the energy cost of inference at the edge

One of the players at the forefront of this technology is Dutch startup Innatera. Fresh from speaking on stage at the 2025 Hello Tomorrow Summit, CEO Sumeet Kumar sat down to share some thoughts.

“For AI to impact our lives, it must leave data centers and enter the world around us. This requires 10,000x efficiency gains. This is only achievable through brain-inspired architectures.”

What we need is this:

– Ultra-low power consumption for use in battery-constrained environments

– Real-time sensory data processing at the edge

– Adaptive learning capabilities

Taking inspiration from the brain is nothing new: machine learning is based on neural networks modelled on the brain. But this is a fairly loose resemblance.

Neuromorphic chips, on the other hand, directly mimic the structure and function of real neural systems, using spiking neural networks (SNNs) to process information in discrete events, just like the brain’s neuronal firing patterns.

In this way, they allow sensor data processing at the source, currently reducing latency by up to 100× and energy consumption by up to 500×. These characteristics make it possible to efficiently process the world’s sensor data as soon as it is captured, at the edge – essential for applications such as predictive health monitoring in wearables.

Tackling data movement: Bringing storage closer to compute to reduce latency and energy use

Data movement is another element that uses enormous amounts of time, energy, and cost in AI workloads.

Weebit Nano is using neuromorphic computing to tackle this.

In collaboration with CEA-Leti (Laboratoire d’électronique des technologies de l’information – France), they are developing ReRAM (Resistive random-access memory), a non-volatile memory technology that is faster, more energy-efficient, and scalable.

The major advantage of this over traditional technologies is that it’s manufactured at scales below 10 nanometers. In other words, the electronic “roads” inside the chip are much shorter, bringing the benefits of it being faster and more energy-efficient than traditional chip architecture.

By embedding ReRAM directly on the chip, neural network weights can be stored locally, reducing the need for external memory and enhancing performance in AI accelerators and microcontrollers.

Microfluidics to cool AI chips

To build on the new deep tech solutions and maximise performance, microfluidics is being introduced as a way to enhance the thermal exchange and unlock the real potential of novel AI chips.

Tiny channels etched onto the back of the silicon chip bring liquid coolant directly inside the silicon and more efficiently remove heat. This removes the need for more traditional cooling plates, which have limited impact in reducing the energy consumption of data centers. Microsoft recently partnered with Corintis to develop this, emphasising the value of corporate–startup collaboration.

Overcoming scaling hurdles to build European sovereignty

Given the application of AI across all industries, chips and semiconductors will define the future of industry and deep tech.

But inherent scaling challenges remain.

– Fragmented Supply Chains: 75% of semiconductor manufacturing is concentrated in East Asia, creating bottlenecks and dependence.

– Capital Intensity: Innovative chip development requires hundreds of millions of dollars to be deployed at an industrial scale.

– Regulatory Constraints: Export controls and investment restrictions limit cross-border collaboration.

Given the huge concentration of semiconductor manufacturing in East Asia, Europe must become a key player in this space to reduce dependency. This could be the last chance for European sovereignty.

Innatera achieved its first customer-sampling prototype with only €5M, rather than the €150–200M that is typical for AI chips. By working with a growing base of early customers, they were able to de-risk their technology incrementally over multiple prototypes and ultimately enter production with phenomenal capital efficiency.

As outlined in our report ‘Deep Tech Decoded’, startups that reduce the technological risk by reaching new industrialisation milestones create substantial value for investors, even pre-revenue. These milestones include proving viability outside lab conditions and then industrial scale, which bring the company closer to entering an attractive market with the added protection of intellectual property.

Europe must grow its ecosystem to tackle these challenges

Semiconductor sovereignty is recognised by European leaders as an essential component of Europe’s technological and economic competitiveness.

But what does ‘semiconductor sovereignty’ even mean?

It means creating a robust, independent, and flourishing ecosystem for semiconductors and AI chips within Europe, with as little dependence on other nations as possible.

To achieve this goal, it is not enough to simply fund isolated projects. Europe needs to map emerging technologies, invest in foresight, and carry out systematic market analysis to identify where its strengths lie and which strategic collaborations can deliver the greatest impact.

And Europe has already made some good progress in this respect.

The EU Chips Act in September 2023 committed over €43 billion in policy-driven investment through 2030 to boost its semiconductor ecosystem. This covers funding for pilot lines, packaging, design tools, and research centers, to double Europe’s global market share in semiconductors to 20% by 2030.

The EU Startup and Scaleup Strategy targets the specific hurdles that early-stage tech firms face in Europe, offering concrete measures such as a more integrated venture capital market, simplified cross-border rules, and harmonised infrastructure access to help scale semiconductor and AI-related startups within Europe, rather than lose them to overseas ecosystems.

Together, the European Commission gives European AI chip startups access to capital, connections with industrial partners, and strategic collaborations to become a leader in Europe.

Mistral leading the charge in Europe

On the topic of sovereignty and reducing reliance on foreign powers, in September, we saw possibly the biggest move yet.

ASML, the Dutch superpower developing advanced photolithography systems, became the majority shareholder in the French AI unicorn Mistral AI, with a €1.3 billion investment as part of their Series C round. ASML’s equipment allows chip manufacturers to fabricate faster, smaller, and more energy-efficient integrated circuits.

Mistral, for their part, started as an AI software company, but after announcing their move into infrastructure in June in a partnership with NVIDIA, they are quickly becoming one of the most important scaleups in Europe.

This represents a great example of Europe building its independence.

A corporate playbook for the AI compute era

So how should corporates navigate this transformation? The answer lies in early positioning, foresight, and strategic ecosystem engagement. Here are our top tips for corporates:

– Map emerging trends: Identify early-stage breakthroughs in research, closely monitor technologies transitioning from lab to market, and follow where investment capital is flowing.

– Build strategic partnerships: Engage with deep tech startups early, securing access to IP and pilot projects before competitors.

– Invest in resilience: Build supply chain strategies that reduce geographic dependencies.

The endgame is clear: seamless, context-aware computing integrated into daily life. 

As Sumeet envisions:

“Imagine smartwatches predicting diseases before symptoms appear – this requires continuous  processing of health data locally, on-device, with brain-like efficiency.”

The future of semiconductors and AI computing will be defined not just by breakthroughs, but by who can anticipate trends and build the right collaborations to deploy them on a global scale. Successful execution requires connecting deep tech science, startups, and corporates as soon as possible.

Hello Tomorrow Consulting partners with corporations to turn deep tech into real business opportunities, supporting with new market entry, foresight & trends analysis, startup sourcing, corporate venturing strategy, and more.

Interested in working with us? Get in touch. 

Or, if you’re a founder of a deep tech startup, apply to the world’s longest-running startup competition, the Hello Tomorrow Challenge.

Watch the full panel discussion from the 2025 Hello Tomorrow Summit here.