Given the pace of technological changes in the past decades, predicting trends may seem like a fool’s errand. But we think it’s important to have a perspective, so with no further ado, here are our predictions about what forces may continue to shape and reshape our industry in the year(s) to come.
1. Helping machines “think” better
Recent years have seen huge steps forward in machine learning, deep learning, and artificial intelligence (AI). Much of the focus has been on training the models upon which AI services are built; there will now be a shift in momentum from training to inference.
Inference is more akin to thinking and reasoning, where trained models are applied to data to derive predictions and conclusions. Chip technologies suited to “thinking” over “learning”, and therefore leading to more accurate outcomes from AI, will come to the fore. There’ll also be increased focus on neural processing units (NPUs) over GPUs, and particularly where these are embedded close to the source of data (see our next trend for more).
Advances in photonic integrated circuits (PINs), which increase the speed and capacity of data transfer, will enable ultrafast artificial neural networks and power neuromorphic computing. Specific to computer vision applications, convolutional neural networks (CNNs) will transform the ability of machines to interpret and understand images and visual information.
2. More AI moves to the edge
The continued innovation in AI chip technologies – not least in relation to the development of NPUs – will unlock a growing opportunity to add more intelligence to every connected device and sensor. This is edge AI, where AI capabilities are embedded in devices that sit at the very edge of the network.
Placing intelligent analytics within devices and sensors – and therefore much closer to the source input data – reduces latency and increases data security and privacy. AI processing at the edge also reduces the data sent across the network from the device to the data center, with an associated reduction in processing needed on servers. NPUs using lighter AI models such as TinyML are also far more energy efficient than GPUs found in data centers.
When we consider the number and type of connected devices and sensors, the opportunity for edge AI across every industrial and consumer sector is obvious. Innovation in this area, adding “intelligence” to every device, will be rapid.
3. New approaches to silicon
Ways to increase the power of semiconductors in more efficient ways will be a priority. Silicon Carbide (SiC) is a perfect example. It’s properties and benefits for power electronics are already well-known, and its potential in automotive, energy, and industrial applications is huge.
However, SiC semiconductor manufacturing is challenging. Innovations in this sector will come through closer vertical integration of the manufacturing process, from design to test, resulting in improved yields and quality.
Silicon photonics has also emerged as a technology ideally suited to some of today’s, and tomorrow’s, compute challenges. Using light (photons) rather than electrons to transfer information, silicon photonics increases the efficiency of data transfer with much lower latency than traditional electronic semiconductors. This makes it an ideal technology for interconnects within AI data centers, but the potential applications are almost unlimited.
4. Traditional chip technology drives Quantum forwards
Quantum computing has been discussed conceptually for years, though might have seemed more a work of science fiction that having real world applications. The step forward in computing power that quantum represents is so great that it suggests a completely new approach to processing technologies is needed. However, with just a relatively small adaptation, established semiconductor manufacturing technologies can be used to power quantum computers. This will change the game.
The use of proven FD-SOI semiconductor process technologies will accelerate quantum’s development towards real-world applications. While quantum isn’t suited to every computational task, we’ll see exploration of potential use cases across every industry sector and application, from financial services to pharmaceutical, from cybersecurity to climate modelling.
5. Biosensors move from fitness fanatics to everyday healthcare
Millions of amateur athletes already diligently use wearable devices to monitor bioindicators during activity and measure their progress towards improved fitness.
Advances in biosensors – the number and type of bioindicators tracked, reduced size and cost, and vastly improved power efficiency – will see them embedded in a greater variety of devices and materials. When balanced with control regarding what to monitor, who to share that information with, and when, people will feel comfortable about ongoing monitoring of their health indicators.
The use of biosensors will move beyond the worlds of personal and professional fitness, and into the broader provision of healthcare services. In combination with edge AI, medical advice and diagnosis will be available when needed, and often without requiring a trip to a surgery or clinic. Proactive healthcare – prevention rather than cure – will have never been more possible and has the potential to remove a significant burden from the global healthcare system.
6. EVs back on track, smarter and safer than ever
While there’s no question that in some markets sales of EVs have been in decline, and that overall rates of sales growth have been lower than predicted, on a global basis, volumes of EV sales are still increasing.
Sales figures aside, innovation in the capabilities of EVs will continue apace. There’s no doubt that semiconductors, sensors and software will play an ever-growing part of the automotive experience, and safe operation of vehicles.
Consumer adoption of EVs will also be supported by more efficient battery and power management within the EV itself, and more available and efficient charging infrastructure across countries and regions. The commitment of consumers to play a positive role in the transition to renewable energy sources will grow, with EVs being one of the most accessible and impactful steps they can take.
7. Everything has a digital twin
Digital twins – digital representations of physical entities, from machines, to buildings, to entire cities – allow for virtual modelling, testing of planned improvements, and more rapid iteration of real-world enhancements. Fundamental to the creation of a digital twin is a stream of accurate data about the physical entity, typically provided by IoT and edge AI-enabled sensors.
Innovation in intelligent sensors will mean that almost everything physical will be “twinnable” digitally, with the insights gained from data leading to advances in design, monitoring, and process optimization. Almost every type of sensor, from temperature to pressure, from air quality to sound, has a role to play.
City authorities will use digital twins to reduce pollution, improve transportation, enhance safety, and meet broader sustainability goals. We’ll create digital twins of our own homes, using them to optimize energy usage and as the basis for increased smart home automation. Transportation, hospitals, airports, factories, sports venues…everything will have a digital twin.
8. To infinity, and beyond
We’re in an unprecedented age of placing satellites into space. There are currently around 9,000 satellites in orbit around the earth, but this number is expected to grow to as many as 60,000 by the end of the decade.
Much of this growth will be driven by the “mega-constellations” of satellites in low earth orbit, which are creating low latency, high performance communications networks spanning the globe.
Looking beyond the earth’s orbit, many nations are planning space explorations. It’s likely that within the next few years a human will once again set foot on the moon. One of the key goals of these planned expeditions is to look for and analyse the raw and rare minerals that might unlock the next age of technological innovation.
If recent history tells us anything, the next year (and beyond) will see some startling advances in technology, with the pace of change continuing to accelerate. Some of the above trends might prove to be accurate, others could be wishful thinking on our part. It’s almost certain that innovations will also appear that we’re not yet aware of.
Time will tell, but we’re excited to find out.
- Learn more on artificial intelligence at the Edge
- Discover our latest technology infrastructure for EVs
- Discover our latest biosensors
