Oracle is making a monumental bet on Artificial Intelligence, solidifying its position in the cloud market through an extensive partnership with AMD. This collaboration, featuring next-generation AMD EPYC CPUs and powerful Instinct AI GPUs, is designed to offer unparalleled price-performance, address NVIDIA supply constraints, and power the demanding AI workloads of partners like OpenAI, signaling a significant shift in the competitive landscape for cloud and AI infrastructure.
The landscape of cloud computing and artificial intelligence is undergoing a profound transformation, with tech giants aggressively vying for supremacy. At the forefront of this evolution, Oracle (ORCL) is executing a strategic masterstroke, cementing a deep and expansive partnership with Advanced Micro Devices (AMD) that promises to redefine its cloud offerings and fuel its burgeoning AI ambitions. This alliance is not merely about hardware; it represents Oracle’s calculated move to deliver unparalleled price-performance, navigate supply chain challenges, and secure a pivotal role in powering the next generation of AI workloads.
Oracle Cloud Infrastructure Evolves with AMD EPYC Processors
Oracle Cloud Infrastructure (OCI) is making significant strides in optimizing its cloud setups by adopting AMD EPYC processors. The introduction of new compute e5 instances, featuring the latest 4th generation AMD EPYC processors, underscores Oracle’s commitment to striking a better balance between cost and performance for its clients. Unlike the rigid, often wasteful instance options offered by other cloud providers, OCI’s flexible instances empower customers to precisely allocate cores and memory according to their specific requirements, thereby significantly reducing costs associated with unused computing resources.
This flexible approach has already yielded substantial benefits, with Oracle noting that OCI customers utilizing flexible instances with AMD processors are collectively saving an impressive $40 million annually, as stated by Donald Lu, Senior Vice President, Software Development, Oracle Cloud Infrastructure. These new e5 instances offer:
- More CPU cores and better performance per core.
- Improved memory bandwidth.
- Higher storage capacity compared to previous iterations.
According to internal testing by Oracle, the new e5 standard instances deliver 33% better performance per core and 50% better memory bandwidth, while HPC instances offer 40% better price-performance, showcasing a clear generational leap. These advancements enable faster and more efficient handling of diverse workloads, including web servers, enterprise applications, and complex mathematical simulations, supporting various operating systems and installable images from the Oracle Marketplace.
The AI Imperative: Oracle’s Bold Generative AI Bet
Recognizing Artificial Intelligence as the transformative future of the technology industry, Oracle has aggressively pivoted towards integrating generative AI capabilities across its application lineup. A notable collaboration with Cohere, a Canadian AI software firm, highlights Oracle’s intention to enhance its offerings with advanced AI functionalities. This move positions Oracle squarely in the competitive arena alongside established players like OpenAI and Anthropic.
Oracle’s AI strategy is distinct, primarily focusing on leveraging AI within its cloud infrastructure tools. This contrasts with Meta Platforms (META), which prioritizes AI for data management, Alphabet (GOOGL), which integrates AI into search engine capabilities, and Amazon (AMZN), which uses AI to enhance its web services. This focused investment in AI, coupled with its robust cloud performance, has significantly boosted Oracle’s market standing. Shares of Oracle have gained 42.7% year-to-date, outperforming the Zacks Computer and Technology sector. The Zacks consensus estimate for Oracle’s first-quarter fiscal 2024 earnings projected a profit of $1.17 per share, signaling year-over-year growth of 13.59%, with revenues expected to hit $12.4 billion, an 8.33% increase.
AMD’s Ascent in the AI Chip Market: Challenging NVIDIA
The surging demand for AI infrastructure, coupled with reported NVIDIA (NVDA) GPU shortages, has created a fertile ground for AMD to expand its footprint in the high-performance computing and AI accelerator markets. Oracle is strategically deploying AMD’s Instinct MI series GPUs as a powerful alternative. Initially, Oracle is set to deploy AMD Instinct MI300X AI and HPC GPUs for its cloud services, with deployment anticipated in early 2024, as reported by Reuters.
Looking further ahead, Oracle Cloud Infrastructure plans a massive deployment of 50,000 advanced AMD Instinct MI450 chips, commencing in the second half of 2026, with further expansion in 2027 and beyond. These MI450 chips are a game-changer for AMD, being the first AI chips from the company that can be combined into a larger rack-sized system, enabling 72 chips to function as one to create and deploy cutting-edge AI algorithms.
Unveiling the AMD Instinct MI355X: Powering Next-Gen AI
In a recent development, Oracle and AMD announced the availability of AMD Instinct MI355X GPUs on OCI. These processors are engineered to deliver breakthrough performance for large-scale AI training and inference workloads, offering over 2x better price-performance compared to the previous generation. OCI plans to offer Zettascale AI clusters accelerated by up to 131,072 MI355X GPUs. Key technical specifications and benefits of the MI355X on OCI include:
- Significant Performance Boost: Up to 2.8x higher throughput, enabling faster results and lower latency for larger AI workloads.
- Larger, Faster Memory: 288 gigabytes of HBM3 and up to eight terabytes per second of memory bandwidth, crucial for executing large models entirely in memory.
- New FP4 Support: Support for 4-bit floating point compute (FP4) standard, allowing cost-effective deployment of modern large language models and generative AI.
- Dense, Liquid-Cooled Design: Maximizes performance density at 125 kilowatts per rack, with 64 GPUs per rack, leading to faster training times.
- Open-Source Stack: Leveraging AMD ROCm, an open software stack that includes popular programming models, tools, compilers, libraries, and runtimes for AI and HPC solution development, ensuring flexibility and preventing vendor lock-in.
- Network Innovation with AMD Pollara: Oracle will be the first to deploy AMD Pollara AI NICs on backend networks, providing advanced RoCE functionality and support for open industry standards from the Ultra Ethernet Consortium (UEC).
Strategic Partnerships: OpenAI and the Future of Compute
The strategic intertwining of Oracle, AMD, and OpenAI underscores a concerted effort to build a resilient and diversified AI computing ecosystem. OpenAI, the creator of ChatGPT, has reportedly inked one of the largest cloud deals ever with Oracle, potentially worth up to $300 billion over five years, signifying a massive commitment to Oracle’s cloud infrastructure. This robust collaboration provides Oracle with a marquee client whose insatiable demand for computing power will drive further innovation and expansion.
Furthermore, OpenAI has also forged a multi-year deal with AMD for processors, which includes an option for OpenAI to purchase up to roughly a 10% stake in the chipmaker. OpenAI is actively working with AMD to refine the design of its MI450 chips for AI workloads and plans to construct a one-gigawatt facility based on these processors. This strategy by OpenAI to seek computing power from multiple sources, even exploring designing its own AI chips with Broadcom, reflects a broader industry trend towards diversifying supply chains and reducing dependency on a single vendor.
Investor Outlook: What This Means for Oracle and AMD
For long-term investors, the deepening alliance between Oracle and AMD presents a compelling narrative of strategic growth and market disruption. Oracle’s aggressive expansion in cloud infrastructure, coupled with its focused investment in generative AI and flexible instance offerings, positions it as an increasingly formidable player in the cloud wars. The company’s ability to attract and serve high-profile AI workloads, exemplified by its deal with OpenAI, solidifies its revenue growth prospects and market share expansion.
For AMD, this partnership is a significant validation of its Instinct GPU lineup and EPYC CPU dominance in the data center. Securing major contracts with a cloud giant like Oracle, especially as an alternative to NVIDIA, signals AMD’s growing prowess and market acceptance in the crucial AI accelerator space. The sheer scale of planned deployments, including 50,000 MI450 chips and extensive MI355X clusters, translates into substantial future revenue for AMD and strengthens its competitive standing against NVIDIA. Investors should closely monitor AMD’s execution on these large-scale deployments and its continued innovation in AI hardware and software, particularly with its open-source ROCm platform, as these factors will be critical determinants of its long-term success.
The Oracle-AMD alliance is not merely a transactional agreement; it represents a foundational shift in the AI infrastructure landscape. As demand for AI compute capacity continues to surge, this partnership promises to deliver high-performance, cost-effective, and flexible solutions, setting the stage for sustained growth and innovation for both tech giants in the years to come.
