Author: Liu Wenxuan.
Generative artificial intelligence (GenAI) runs through the hot topics in 2023, from the very beginning of ChatGPT to the emergence of industry large models and device-side models, GenAI will also leapfrog into 2024 and continue to be the focus of the industry.
Not long ago, at the Dell Technologies 2024 Technology Communication, John Roese, Global Chief Technology Officer of Dell Technologies, shared with us the emerging trends affecting the technology industry in 2024 and beyond. He said: "This year is the first year of GenAI, especially with the rise and emergence of ChatGPT and other GenAI systems, and now the whole industry is discussing and trying to understand GenAI. But it's important to note that GenAI, or AI as a whole, will continue to dominate the conversation in 2024. ”
John Roese, Global Chief Technology Officer, Dell Technologies
John Roese said he has four major outlooks for 2024 and beyond in the technology industry, including GenAI, zero trust, edge platforms, and quantum computing.
"AI is at the center of the world of the future, where it can be put into production at the edge, secured with zero trust, and ultimately powered by quantum technology to deliver the performance and efficiency needed to scale to global systems," said John Roese. We need to keep a close eye on AI, but we can't ignore other architectures either. This is the only way to ensure long-term success by aligning vision with action. ”
Generative AI is further grounded
In recent years, the digital world is constantly changing and developing at an unprecedented speed, and as a booster of the industry, the topic of AI has further become the focus of the entire industry with the emergence of GenAI such as ChatGPT. "If this year is the first year of GenAI, 2024 will be the year after generative AI, and GenAI will continue to dominate the conversation and move from theory to practice," said John Roese. Shift from training infrastructure and costs to inference and operational costs. Increasingly, the focus will shift from broad experimentation to a top-down strategic focus, and the single-out handful of truly transformative GenAI projects will be held accountable by leaders. ”
"While GenAI has sparked tremendous creativity in how it can transform business and the world, there are very few GenAI events at scale in the real world," said John Roese. Entering 2024, we will see the first batch of GENAI enterprise projects mature. ”
For enterprises in various enterprises and different vertical industries, whether it is GenAI or GPT4, the training cost of large language models is very high. However, businesses in vertical industries often don't use these models directly in their applications. Therefore, there are a number of issues that must be addressed to further develop these models so that they can be truly applied to the production environment of vertical industries and be truly productive. This includes the establishment of an inference infrastructure, which businesses must have a better understanding in 2024. In addition to training infrastructure, you also need to think about how to build an effective inference infrastructure. In addition, you need to determine where your inference capabilities are deployed. In the past, enterprises have primarily placed training in the data center, but in the future, careful consideration must be given to placing inference at the edge, close to the call center or customer location, and this must be made clear.
With the continuous iteration of GenAI, the concept of AI PC has also begun to enter the public eye. In fact, the development of AIPC has been around for a while, and the essence of AI PC is still a PC platform, but it is optimized for AI-based ** in design. He predicts that 2024 will be the early days of AI PCs, and by 2025 and 2026, most of the world's PCs will add additional processing power, and the applications that can be supported will not only be Copilot, but any web browser-based will be supported. "There will be more and more use cases for AI, but there will be many AI use cases that can run directly on PCs in the future, instead of running in the cloud, because AI will become more and more mature and powerful," he said. ”
Zero trust continues to gain traction
"Security" is always an important issue that must be faced in the digital age, and even more so in the era of GenAI. As an innovative security concept in the digital era, zero trust will also be further implemented at the level of technology and standards by 2024. "2023 was filled with discussions about zero trust and its importance in global cybersecurity efforts, and in 2024, zero trust is no longer just a hot topic," said John Roese. We're going to focus more on implementing real technologies, standards, and even certifications to clarify what true zero trust is. ”
Zero trust has significant value, but it also requires recognizing the challenges of deploying a zero trust infrastructure in legacy environments. Zero trust infrastructure is easiest to implement in a greenfield environment because in these edge platforms, zero trust principles can be applied from the start. However, much of the current AI infrastructure is still under construction, so we expect zero trust security principles to become the default choice in building AI infrastructure. Five years from now, any IT architecture that doesn't adopt zero trust will be out of date.
Edge platforms continue to thrive
The rapid growth of edge platforms continues. Although cloud service providers rely primarily on centralized data centers to provide services in a multi-cloud environment, in real life, data does not exist only in data centers, but is distributed across edge devices such as factories and hospitals. As a result, we need to store data in scattered locations, which means that large service providers need to build their own edge platforms for each edge device. However, it is clearly not ideal for a company to use five different edge platforms to provide services at the same time.
According to John Roese, enterprises will recognize two ways to build a modern edge – the expansion of edge silos and the building of multi-cloud edge platforms. For the future, the direction will be the latter, with the adoption of an "edge platform" to make the modern edge an extension of a multi-cloud infrastructure.
Once the edge platform and edge workloads are separated, the infrastructure can support us in building a parallel platform. This platform will include software and hardware orchestration to operate on a software-defined platform. For cloud services, instead of having a separate platform for each workload, we can use a shared platform. Dell Technologies believes it is possible to further improve security and automation by creating a shared edge platform instead of creating a different platform for each workload and application. At the same time, Dell Technologies has set a plan to begin realizing this vision in 2023 and will begin to implement some of the previous concepts and assumptions next year. Since many enterprises need to process data outside of the data center, when leveraging GenAI, we need to operate on an edge platform to realize the possibilities of GenAI when using data.
Quantum computing and genai continue to drive the future
According to John Roese, the advent of quantum computing solves the major problem of the enormous demand for computing resources in GenAI and many large-scale AIs. Quantum computing is expected to bring about a huge technological leap forward and improve the capabilities of AI systems. Modern AI will be built on the computational foundation of hybrid quantum systems, and AI work will be distributed across multiple computing architectures, including quantum processing units.
John Roese argues that quantum computing and GenAI are inextricably linked, and that significant changes could occur at any time in the next five years. He said that unlike traditional computing, quantum computing is a probabilistic way of computation, and if genai can be optimized, it will make it more efficient. This increase in efficiency will be unimaginably huge, just like the surprise that ChatGPT brought to people when it came into being.
Despite the many challenges, John Roese is confident in Dell Technologies' ability to lead customers forward by being a strategic partner through collaboration and mentorship. He stressed the need to build a digital continuum that works together to turn digital ideas into reality. Dell Technologies is an integrated company at the forefront of technology and business growth opportunities, with the expertise, solutions and partners to help customers build a data-driven future.