Since the dawn of the 20s, the technology space has been largely dominated by one thing – AI. The demand for AI and AI applications has driven computational power growth to new heights, with tools such as Generative AI leading the charge.
However, this new landscape comes with its own set of issues for organisations, namely scaling the growth of AI, managing computational growth, efficient power distribution and management, navigating component availability, risk mitigation, and predicting what might come next.
To learn more, and how to overcome these challenges, Procurement Pro’s Harry Fowle spoke with Rob Picken, SVP of Digital Transformation at Sourceability.
The driving factors behind computational growth
The demand for computational power is being driven by a surge in the adoption of AI and machine learning technologies across a plethora of industries. For Picken, AI is no longer confined to niche or specialised applications as it had historically been; instead, it is being increasingly deployed to solve a broader range of problems.
Generative AI must be mentioned when considering why. Unlike traditional machine learning, which often deals with specific, well-defined tasks, Generative AI has the capability to tackle open-ended and complex problems. Picken noted: “When you’re looking at Generative AI, that’s to solve big problems, to solve open-ended problems… the instructions are very unformatted questions.” This transition has significantly increased the computational requirements as these AI models need more power to process and generate vast amounts of data.
AI is also being deployed in many everyday products and has become a market buzz. From smartphones and smart displays to cars, AI is being embedded into a wide array of products. This trend, while enhancing the functionality of these products, also drives the need for more computational power. But while the push to incorporate AI into various products is strong, Picken warned that this trend might sometimes be driven more by the buzz around AI than by genuine need. This could lead to inefficiencies and potential issues in the future as companies may be investing in AI technologies that are not essential to their core operations.
As AI models grow in complexity, their energy consumption also rises significantly – something that is already being seen with each model of ChatGPT, for example. A single AI-driven search query can consume vastly more energy than a standard search. This increase in energy demand is a critical factor in the growth of computational power, and it also raises concerns about the sustainability and environmental impact of AI technologies. Picken provided a stark example: “It’s 15x as much energy used when you’re asking an AI to search compared with a normal Google search.”
Managing computational growth
As organisations increasingly adopt AI and other computationally intensive technologies, managing this growth has become a critical concern. The challenge lies in balancing the opportunities presented by these technologies with the practical realities of implementation, resource allocation, and sustainability.
One of the key factors which Picken emphasised here is the importance of focusing on core business needs when integrating AI. Whilst AI can offer exciting possibilities, not every business process or product requires AI to be effective or reach potential. “Users and industries have to look at whatever they’re doing with AI doesn’t escape the fact they have to produce things that consumers and companies want to buy.”
Rather, businesses should strategically deploy AI in areas where it can provide the most significant benefits. This could involve using AI to enhance decision-making processes, improve customer services, or optimise operations. The key is to ensure that AI applications align with the company’s goals and deliver measurable outcomes. Picken pointed out that while some companies like Nvidia are thriving, there are still questions about when and how other businesses will start seeing substantial returns from their AI investments: “When is any company going to make actual money from AI? So that’s the race right now.”
How AI is impacting component demand
The growth of AI is not happening in isolation; it competes with other significant technological advancements for resources, particularly in areas like the electrification of vehicles and power grids. The demand for components such as high-density memory and microprocessors, which are essential for AI, is also critical for other industries, creating a competitive market for these resources.
Picken explained: “High-density memory, which is extremely critical for AI applications, is also critical for automotive applications. So, at the same time as organisations are cramming AI into everything, these same components are in just as much demand elsewhere.”
This is also the case in power conversion and energy storage. Efficient power conversion and energy storage solutions are vital for managing the energy needs of AI systems. Power conversion technologies, such as DC-DC converters, are essential for ensuring that AI hardware receives the appropriate voltage and current levels. Energy storage solutions, including advanced batteries and capacitors, are also crucial for managing peak power demands and ensuring reliable operation, creating a large array of competition for components.
The increasing demand for AI-related components directly adds strain to global supply chains, which are already under pressure from other sectors. The competition for these resources can lead to shortages, price increases, and supply chain disruptions, making it more difficult for companies to secure the components they need.
Mitigating the risk
Whilst this new landscape poses new challenges, organisations can certainly take measures to ensure that they do not fall victim to supply chain challenges. Here, Picken offers his insights into what companies can do to mitigate risk.
For starters, organisations should identify which components are critical to their operations, particularly those with limited availability due to external factors like temperature ranges, vibration resistance, or specific industry regulations (e.g., medical devices, military, aerospace). Understanding these requirements allows companies to prioritise these components in their supply chain strategies. “If you’re in a particularly high-reliability industry – medical devices, military, aerospace, EV charging infrastructure, power grid generation distribution – there’s actually only certain components that you can use because of either external qualification or temperature ranges or vibration resistance or radiation hardening, for example.”
Companies should also begin to implement a triage process to assess the availability and risk associated with each component, implores Picken. This process involves evaluating the criticality of components, checking their availability, and determining if alternatives can be used. By focusing on the most crucial components first, organisations can reduce the risk of supply chain disruptions.
Organisations can also take steps to leverage data tools and work with experienced account managers to gain better visibility into the market and understand the complexities around component availability. “You can check [component] availability with systems like DataLink, which is a Sourceability proprietary programme, or you could speak to one of the experienced account managers that we have here to understand what the market complexity is around those parts.” This can help organisations to make informed decisions or provide real-time insights.
For components that are less critical, companies should explore the possibility of using alternative parts. Qualifying alternate components in advance ensures that if a shortage occurs, the organisation can quickly switch to a substitute without any significant disruptions.
Diversifying suppliers, increasing inventory for critical components, and collaborating with independent distributors can also help to build a more resilient supply chain. Companies should not put themselves into a position where they are relying solely on a main, or a handful of, suppliers, as this approach can leave them vulnerable to shortages. Picken advises: “Where can you then go and get information from independent distributors you might ask? Sourceability, as an example, is a non-biased, agnostic source of information and can be a great source of data.
“If you just rely upon the suppliers who provide you with 80% of your components, then you will always be at risk of losing focus on that other 20%.”
Finally, whilst it may sound obvious, one of the best practices an organisation can perform is to stay informed about potential supply chain risks and market changes. This includes monitoring geopolitical developments, environmental factors, and industry trends that could impact component availability. Staying vigilant allows companies to anticipate issues and adjust strategies accordingly. After all, as Picken suggests: “By making sure that you have full visibility of everything, that’s how you can start to plan more effectively and make better decisions.”
This article originally appeared in the September issue of Procurement Pro.
