Power, Partnerships, and the unspoken Role of Supply Chains: VCA at SEMICON SEA 2025

The first speaker after the opening address by Singapore’s Economic Development Board was Dr. Prabu Raja, President of the Semiconductor Products Group at Applied Materials. His presentation, titled “Collaborative Innovation – Connecting Globally to Solve the Challenges of the Future,” immediately set the tone.He opened with a look at the semiconductor industry’s growth, from the invention of the transistor 65 years ago to today’s $600 billion semiconductor sector. His projection that this figure is expected to double in the next six years, thanks largely to the accelerating demands of AI.However, Dr Raja pointed out that the computational needs of AI are causing an exponential increase in energy consumption, with data centers alone consuming 400 terawatt hours, projected to grow to 950 terawatt hours in the next six to seven years. To put it into perspective, he noted that this figure would be equivalent to 18 times Singapore’s entire annual electricity usage. His call to action was clear: the semiconductor industry must radically improve energy efficiency, aiming for another 10,000-fold improvement in energy efficiency in the next 15 years. He emphasized that this isn’t just about faster chips but optimizing for power per watt.What was also a reality check, hastening the call for partnerships and collaborations within the semiconductor ecosystem, whether it is vendors, suppliers, customers and manufacturers, to ensure that the issue of long commercialization timelines, which can take 10 to 15 years, is not an obstacle. He stressed the need to innovate the way the players innovate to address complexity and speed.

The second speaker, Tim Breen, CEO of GlobalFoundries, picked up where Dr. Raja left off. His talk, “Partnerships in a Changing World,” echoed many of the same concerns but framed them with a sharper lens on global connectivity and collaboration. Breen reiterated the 950 TWh forecast was enough electricity to charge 80 trillion iPhones. It was a figure that made the abstract very real. Yet, even with all this hardware capacity, Breen noted that many GPUs today are sitting idle. This is not due to lack of processing power, but because data connectivity is failing to keep up. He emphasized that this is where silicon photonics comes in. It is a technology that could help solve the data bottleneck and allow the hardware to perform at full capacity.Beyond the tech, Breen encouraged thinking about the role of partnerships and collaboration in driving innovation, particularly in areas like AI, robotics, and bioscience. With more regulatory constraints, complex IP landscapes, and geopolitical tensions, how do companies continue to collaborate meaningfully across borders? He also touched on the future of talent, including the integration of AI coworkers in companies within the next five years. He suggests rethinking the role of HR in managing a workforce that includes non-human talent.

Finally, what I called the lightning bolt statement offered by the third speaker, Dr. Rutger WIJBURG, Chief Operations Officer, Infineon Technologies, Germany was what nailed the session for us. Somewhere in the middle of his densely packed slides, he shared a line that cut through the technicalities: that at our current trajectory, data centers could consume the entire global energy output by the end of the next decade. It was a statement delivered almost offhand, but its implications were staggering. What struck me was not just the clarity of his message, that the industry is fully aware of its energy challenges and is actively working to address them, but the void it revealed on the other side. While the private sector is racing ahead to reduce consumption and increase efficiency, it’s far less clear who is tackling the question of supply. Are governments stepping in to rethink energy production at the scale needed? Are businesses prepared to invest in the infrastructure necessary to power this digital revolution?From a supply chain perspective, the analog world looms large behind all this digital ambition. Yes, we are building smarter chips and better software. But the materials that make these advances possible, the rare earths, the highly specialized machinery, are not freely or equally available to the businesses that need them.Access is uneven, and the systems supporting them are often fragile. Photolithography tools, for example, are manufactured by only a few players globally. Even if the digital realm races ahead, the analog foundations must still be in place. However, they are often the bottleneck.

This brings us to an uncomfortable but necessary question: While we can read the headline number of 950 terawatt consumption and how our AI and data habits will suck up all the available energy on our tiny planet, I wonder who is addressing how these energy needs will be met? Someone — somewhere — has to generate that power. Will we see a new way of nuclear powered countries or a quicker return to nuclear energy? What about the infrastructure required for the efficient transmission of power throughout the sites of demand? Power generated can only be power available if it is supported by the very analog infrastructure that needs to be built. Or would the next new country to be declared sovereign be an island with nothing but data centres, governed by the companies who run them?It is a strange image we are painting, but not entirely implausible. If the last two decades were defined by the digital divide, then the next two may very well be shaped by an energy divide. And the consequences of that may be just as profound.