Last week’s announcement of Stargate—a $500 billion collaboration between SoftBank, Oracle, and OpenAI—got me thinking deeply about something even more critical than the technology itself: the energy required for such massive artificial intelligence projects. This was top of mind even before this week’s announcement of DeepSeek, China’s alternative AI system, and I touched on it during the 1/24/2025 Post & Walters Ask Us Anything live podcast. When asked about my most urgent priority, I said “Energy,” and for good reason.
Oracle’s construction of massive new data centers highlights just how much space and energy these infrastructures demand. We’re now at a point where nuclear energy is being seriously considered—and in some cases, implemented—as a power source for data centers. While this level of energy use isn’t inherently bad, it raises an important question: How can we consume energy more economically and use it more efficiently?
To put this into perspective, let’s draw a parallel with the evolution of personal computers and operating systems. In the early days, when CPUs, RAM, and storage were expensive, operating system developers worked hard to optimize resource usage. However, as these components became cheaper and more abundant, efficiency often fell by the wayside, giving rise to “bloatware.” For example, the size of operating systems ballooned from Windows 3.1, which required just 6.7 MB, to modern versions occupying several gigabytes. Similar inefficiencies can be seen in software like Optical Character Recognition (OCR) tools.
Now consider Stargate, which relies on an immense data center infrastructure and consumes vast amounts of energy. Just days later, China announced DeepSeek, a competing AI system built at 1/1000th the cost ($5.6 million vs. $500 billion) and designed to use significantly less energy. According to this article, DeepSeek achieves remarkable energy efficiency, partly because sanctions on chip exports forced its creators to innovate rather than rely on resource-intensive “bloatware.” Early reviews suggest DeepSeek is an impressive competitor to ChatGPT, showing what can be achieved through intelligent, resource-conscious design.
This focus on efficiency resonates with my work on energy-conscious innovations at the TWAIN Working Group (TWG). It’s thrilling to collaborate on designing intelligent AI systems while having a meaningful, positive impact on our planet—a trifecta that aligns with my professional and personal values.
One of TWG’s key initiatives is our integration of TWAIN Direct and PDF/R technologies into RISC-V CPU chips. This enables document scanners to operate independently of external computers, reducing the need for additional CPUs, RAM, and energy. With these technologies embedded directly into IoT devices, we can develop smarter systems that consume less power. For instance, machine learning models can train in the cloud to recognize document patterns and classifications, then push those models down to the RISC-V chip for on-device classification. This approach minimizes data transmission to the cloud and enhances energy efficiency, while continuously improving through feedback loops.
Another exciting project is the addition of JPEG-XL to the PDF/R specification. Our recent white paper, “The Benefits of Adding JPEG-XL to the ISO PDF Standard and PDF/Raster,” highlights how this advanced compression format reduces file sizes, conserves storage, streamlines transmission, and encourages digital collaboration over print.
These initiatives are part of a broader strategic objective: providing secure content collaboration solutions. While this challenge might seem solved, it’s an ongoing process of innovation to combat cyber threats and ensure content authenticity, provenance, and transparency. TWG has developed reference platforms for secure documents, integrating blockchain hashing for document verification, C2PA standards for content authenticity, and advanced access controls with biometric and multifactor authentication. These solutions span various types of digital content, from photos and videos to PDF files produced by TWAIN Direct scanners.
Putting all of this together offers us all the promise that A.I. can provide, encourages safe, secure and authentic digital collaboration and, which is of great importance, converses energy so that we can all live our best lives and provide a healthy planet Earth for future generations.
With so many groundbreaking projects underway, the TWAIN Working Group and P3iD Technologies welcome anyone interested to join us. Whether you want to dip your toes in or dive in head-first, there’s a place for you in this journey of innovation.
