Published: 25th April 2024

Perhaps the most groundbreaking change in technology since the birth of the internet is generative AI. Rarely does software have an impact on the business world as much as machine learning algorithms have. Always at the forefront of exploration of the frontiers of tech, the APJ region is already pushing the boundaries of what AI can offer: in media, business intelligence, data processing, marketing, engineering and a hundred other areas.

In one of the largest and widest-reaching surveys on IT in recent years, the Innovation Catalysts study quizzed over 6,000 respondents globally, who gave their answers to a range of queries around innovation, AI, and ML and how their organizations were responding to new technology. (The full survey from Dell Technologies is available here.) The majority of IT professionals (85%) agreed with the proposition that AI and GenAI will significantly transform their industry. And 76% of respondents reported that their organization is already providing intelligent technology in the form of AI optimization software that improves their work experiences.

Given the obvious benefits of AI to all business functions, it’s important, therefore, to understand that access to the compute power and tools required for advanced intelligent algorithms is a prerequisite for today’s businesses. In today’s technology landscape, access to the most advanced AI tools and capabilities can be challenging, especially for businesses looking to innovate and differentiate themselves. The dominance of a few major technology providers has led to a proprietary approach, where the latest AI innovations may be tightly controlled and not easily accessible to a broader range of organizations. This proprietary nature of some AI ecosystems can present obstacles for businesses, especially those in the APJ region, to truly innovate and be creative with AI. They want to use the latest and greatest AI capabilities, but the lack of openness and compatibility between different AI systems gets in their way.

Source: Shutterstock

The need for openness

Developers need open standards to create new uses for AI because it gives them the flexibility to deploy their solutions on-premise, in the cloud, and on edge devices; wherever, in fact, the business’s needs dictate. In parallel with that is the need for compute engines optimized for different devices, capable of delivering AI performance at the point of consumption. Open standards in software and hardware enable interoperability, providing customers the freedom to leverage the AI tools and infrastructure that best suit their unique needs and workflows. This empowers businesses to innovate with generative AI on their own terms, without being limited by a single vendor’s ecosystem.

What the technology industry must pursue, therefore, is a policy of the democratization of GenAI, and those goals are realized by open ecosystems and silicon diversity. Organizations pursuing this strategy of flexibility and choice will gain a significant strategic edge over competitors who primarily rely on public cloud services to manage their AI workloads. This is the key strategy to helping organizations gain a competitive advantage. Empowering businesses with access to the latest hardware optimized for generative AI can further amplify these strategic advantages.

Source: Shutterstock

Powerful hardware

For organizations developing custom AI models and processing large bodies of data, the latest hardware designed from the ground up to be optimized for GenAI significantly lowers TCO, meaning leeway for research and experimentation even within tight budgets.

Better hardware also means projects reach production quicker, and end-users get faster results and an overall better experience. The new Dell PowerEdge XE9680 Server is designed for today’s GenAI workloads. It offers up to eight AMD Instinct MI300X accelerators and provides 1.5TB of coherent GPU accelerator memory per server (the highest ratio in the GPU market currently). That means a lower DC footprint, yet with an increased inference capacity, so very large training datasets can be ingested quickly.

Open software

Hardware power and capabilities unlock an organization’s freedom to innovate without cost overrun, but the software running on it has to offer compatibility with existing AI frameworks, libraries, and models for true portability and compatibility.

Without openness, it’s impossible to achieve that portability across platforms, and therefore, AI can’t be considered democratized. The AMD MI300X Instinct Accelerators in the Dell XE9680 Server, which will be ready to ship in May, offers over 21 petaflops of FP16 performance, yet out-of-the-box, run the common standards in data science of PyTorch and TensorFlow, plus natively supports JAX, Open Neural Network Xchange (ONNX), and OpenAI Triton, inside the AMD ROCm software stack.

ROCm consists of a collection of drivers, development tools, and APIs that enable GPU programming from low-level kernel to end-user applications, and brings together hardware and software optimized for GenAI, large models and fast time-to-market for a business’s AI projects.

AMD’s ROCm is optimized for Instinct MI300X accelerators and is a freely available open software stack that’s capable of evolution and adaptation according to a business’s evolving needs.

An integral part of democratized GenAI is, of course, the open-source ethos. OSS (open-source software) drives quality and excellence, with thousands of users and developers refining and improving the code, allowing increased innovation.

Source: Shutterstock

Better together

Open-source also equates to open flexibility, a situation that means developers can create GenAI-based products and services that operate on a range of devices with upstream support that is available from hundreds of the open-source projects that dominate the GenAI world.

The ultimate flexibility possible today is provided by the combination of the Dell PowerEdge XE9680, AMD’s Instinct MI300X accelerators’ 3rd Gen AMD CDNA (Compute DNA), and ROCm 6 software. A firm foundation and open portability allow businesses and organizations the tools and infrastructure needed to innovate at this critical juncture in technology’s evolution.

GenAI’s transformative powers offer APJ businesses a unique opportunity to develop the next generation of AI-powered software outside the constraints and deliberate roadblocks placed by big tech’s policies of separation and compartmentalization. The horizons can open with just a straightforward deployment configuration tailored to their needs and running on optimized hardware.

To find out more about the Dell PowerEdge XE9680, head to these pages according to your geography: Australia, New Zealand, Singapore, or India. Plus, you can head here to read more about the AMD Instinct MI300X accelerator at the heart of Dell’s next-gen AI-focused hardware.