As the race to adopt artificial intelligence (AI) reaches a fever pitch among businesses, the smartest organizations are already considering how to achieve artificial consciousness – the pinnacle of technological and theoretical exploration. But the effort requires unprecedented hardware and software capabilities, and even as systems are being built, businesses have a long way to go before they understand the demand and even longer before they are able to deploy it. This article is the first of a three-part series outlining the parameters of artificial consciousness.
Hardware requirements include massive amounts of compute, control, and storage. While these enterprise IT categories are not new, the performance requirements are unprecedented. Enterprises are experienced with the compute, control, and storage requirements of Software-as-a-Service (SaaS)-based applications in a mobile-first and cloud-first world, but they are learning how to extend these hardware requirements for AI environments and, ultimately, systems that can achieve the nirvana of artificial consciousness.
It all starts with computing power
As revealed in Lenovo's third annual Global CIO Report, CIOs are currently developing their AI roadmaps, evaluating everything from organizational support to capability building to future-proofing technology investments. The first requirement CIOs must meet when considering artificial consciousness is the computing power involved in capability building. Given the massive amounts of data required to enable a system that is fully capable of learning and inference, the amount of computing required is much more than AI or GenAI.
Advanced processing power is achieved by leveraging a computing fabric made up of advanced server clusters — an approach familiar to CIOs who have deployed high-performance computing (HPC) infrastructure. These clusters seamlessly integrate advanced hardware to deliver unparalleled processing power and efficiency.
At the heart of this cluster-based infrastructure configuration is the concept of a “pod,” carefully configured to maximize computing density and thermal efficiency. Each pod is made up of 16 racks, each housing eight water-cooled servers. This configuration ensures optimal performance as well as environmental sustainability through advanced cooling. These high-performance servers feature 2TB of DDR5 registered DIMM ECC system memory, ensure fast access to data, and combine direct water cooling with a rear-door heat exchanger that captures residual waste heat. These cutting-edge servers can be customized with the latest GPUs or AI processors available from Nvidia, AMD, or Intel to provide massive parallel computing power for this highly demanding application.
Each 16-rack pod also contains Vertiv's end-of-row coolant distribution unit, an innovative component designed to efficiently manage the thermal dynamics of high-density computing environments and ensure this high-performance hardware operates within safe thermal thresholds. The result is a system that delivers high performance and reliability while also being significantly more energy efficient. By reducing overall cooling power requirements, each pod is as powerful as it is environmentally friendly.
Laying the foundations for artificial consciousness
Building artificial consciousness is an ambitious endeavor. To take full advantage of groundbreaking algorithms, entirely new hardware infrastructure requirements will be required. The first of these is computing power. As enterprises scale their processing power, they must also scale their control and storage hardware before they can enable the advanced software stacks and strategic services that operationalize artificial consciousness. In the next article in the series, we'll discuss how to build the capacity to accommodate higher control and storage hardware requirements.
See how Lenovo is unlocking the power of AI for the enterprise.
