JH Explorer in Belgium: Semiconductor industry innovation
Portfolio Manager Richard Clode is excited by cutting-edge innovations unveiled at the ITF World Conference in Antwerp, focused on solutions to address semiconductor power consumption, a key challenge in generative AI development.
2 minute watch
Key takeaways:
- The semiconductor industry’s challenge to reduce power consumption in data centres is intensifying, fuelled by the increasing demands of advanced large language models used in generative AI.
- The industry is making great strides in sustainable energy consumption, via cutting-edge innovation, like enhancements in holistic lithography, backside power delivery and EUV tools.
- This reinforces our enthusiasm for the semiconductor space and its role in driving efficiency in a generative AI future.
| The JH Explorer series follows our investment teams across the globe and shares their on-the-ground research at a country and company level. |
Backside power delivery: a technique of routing power supply lines on the backside of a semiconductor chip or integrated circuit instead of the traditional frontside. This approach increases logic density and improves power and performance.
EUV: Extreme ultraviolet lithography is a type of photolithography that uses extreme ultraviolet light to create intricate patterns onto semiconductor materials, and has advanced by using ever shorter wavelengths.
Gate-all-around transistors: GAA technology is a new semi manufacturing process technology that can help continue silicon scaling, enabling transistors are able to carry more current while staying relatively small.
Generative AI: refers to deep-learning models that train on large volumes of raw data to generate ‘new content’ including text, images, audio and video.
Inferencing: the first phase of machine learning is the training phase where intelligence is developed by recording, storing, and labelling information. In the second phase, the inference engine applies logical rules to the knowledge base to evaluate and analyse new information, which can be used to augment human decision making.
Large language model (LLM): a specialised type of artificial intelligence that has been trained on vast amounts of text to understand existing content and generate original content.
Our ESG integration approach: Thoughtful, practical, research-driven and forward-looking
I’m here at the ITF World Conference, which is one of the leading semiconductor industry events globally, but it’s actually based in Antwerp in Belgium because IMEC, which is one of the leading R&D labs in the semiconductor industry is based here, and has been driving innovation in the semiconductor industry for the past 40 years.
And a key message I’m getting from this conference is that there is one of the largest problems the semiconductor industry has ever had to solve for, which is the exponential growth in the power consumption and data centres required by training and inferencing all of these large language models as we move into a generative AI era.
And you know, the message from these semiconductor companies is, you know, we will solve for this. There’s a huge amount of innovation in their roadmaps and pipeline that will help reduce and flatten that curve of power consumption, like we saw during the internet era, whether that be the shift to gate-all-around transistors to backside power delivery. We’re seeing innovation on the semiconductor equipment side in terms of holistic lithography, embracing machine learning. EUV tools that can have much higher throughput, as well as software innovation. So a full stack solution to this problem, whether it’s algorithms that can reduce the amount of calculus needed to maintain the accuracy of these large language models to be able to do that much faster and therefore, much lower power consumption, or whether on edge devices that can actually inference locally on the device and again, not have to go back to the data centre, which requires a significant amount of extra bandwidth and therefore power consumption. So across the ecosystem, across the supply chain, we’re seeing the semiconductor industry come up with innovative new solutions to address one of the major challenges of our time. And that hopefully should create some great investment opportunities, which is why I’m here today.
Key investment risks:
- The Fund's investments in equities are subject to equity securities risk due to fluctuation of securities values.
- Investments in the Fund involve general investment, currency, RMB currency and conversion, liquidity, hedging, market, economic, political, regulatory, taxation, securities lending related, reverse repurchase transactions related, financial and interest rate risks. In extreme market conditions, you may lose your entire investment.
- The Fund may invest in financial derivatives instruments to reduce risk and to manage the Fund more efficiently. This may involve counterparty, liquidity, leverage, volatility, valuation and over-the-counter transaction risks and the Fund may suffer significant losses.
- The Fund's investments are concentrated in technology sector and may be more volatile and subject to technology related companies risk.
- The Fund may invest in Eurozone and may suffer from Eurozone risk.
- The Fund may charge performance fees. An investor may be subject to such fee even if there is a loss in investment capital.
- Investors should not only base on this document alone to make investment decisions and should read the offering documents including the risk factors for further details.
2 minute watch
Key takeaways:
- The semiconductor industry’s challenge to reduce power consumption in data centres is intensifying, fuelled by the increasing demands of advanced large language models used in generative AI.
- The industry is making great strides in sustainable energy consumption, via cutting-edge innovation, like enhancements in holistic lithography, backside power delivery and EUV tools.
- This reinforces our enthusiasm for the semiconductor space and its role in driving efficiency in a generative AI future.