Advanced Micro Devices (AMD) has used acquisitions to strengthen its artificial intelligence (AI) capabilities and enhance its competitive position. The company has invested over $5 billion to accelerate the development of its Instinct data center GPUs for AI applications. With a current 15% share of the AI hardware market, AMD aims to double this within five years. AMD’s Acquisition strategy centers on providing end-to-end AI solutions, as evidenced by a series of acquisitions over the past three years. AMD has targeted software firms, AI labs, silicon photonics startups, a data center infrastructure provider, and specialized AI teams to increase its share of the $50 billion AI industry.
AMD’s emphasis on open-source software distinguishes it in the competitive AI market. By fostering an accessible and flexible AI software environment, AMD aims to attract a broad range of developers and enterprises. This positions the company as a strong alternative to proprietary ecosystems. These acquisitions signal AMD’s intent to challenge NVIDIA’s dominance in AI hardware by improving software compatibility and performance.
AMD’s recent acquisitions have enhanced its AI capabilities, especially in the rapidly growing data center market. The company’s strategy focuses on delivering robust end-to-end AI solutions, with a key priority on strengthening its open-source software ecosystem. This enables customers to deploy high-performance AI models optimized for AMD hardware efficiently. For example, AMD’s integrated software and hardware can reduce the time from model training to inference to just 48 hours, demonstrating increased efficiency and faster deployment.
Read More: The Digital Gold Rush | Acquiring Online Businesses for Rapid Growth
Key Acquisitions Enhancing AMD AI Prowess
These acquisitions underscore AMD’s deliberate focus on software as a critical component of its strategy. Brium’s technical expertise, in particular, is considered a game-changer for AMD. Ultimately, these acquisitions allow AMD to fuel its AI strategy. They enable AMD to compete directly with Nvidia. The competition is particularly intense in the data center market. This is where there’s the most tremendous potential for revenue growth and profitability.
Untether AI
AMD acquired the team behind Untether AI, a company specializing in AI inference chips recognized for their speed and energy efficiency in edge and enterprise data centers. This acquisition adds a skilled engineering team focused on advancing AMD’s AI compiler and kernel development, as well as digital and SoC design, verification, and product integration. Untether AI’s at-memory architecture is reported to improve performance and reduce power consumption by minimizing data traffic within the chip.
Brium
AMD acquired Brium, a company with leading experts in compiler and AI software, including machine learning, AI inference, and performance optimization. This acquisition strengthens AMD’s ability to deliver optimized AI solutions, particularly for Instinct GPUs. Brium’s expertise in optimizing the inference stack before hardware deployment reduces hardware dependence and enables faster, more efficient AI performance. The Brium team will also support key projects to accelerate open-source tools within AMD’s AI software stack.
Enosemi
The acquisition of Enosemi, a silicon photonics startup, supports the development of photonics and co-packaged optics solutions for next-generation AI systems. Enosemi’s photonic integrated circuits enable faster, more efficient data movement within server racks, which is essential for increasingly complex AI models. Co-packaged optics provide higher bandwidth density and improved power efficiency compared to traditional methods, representing a significant advancement in system architecture. Closer integration between compute and networking is necessary for advanced AI workloads.
ZT Systems
AMD completed a $4.9 billion acquisition of ZT Systems, a data center infrastructure provider, to combine ZT Systems’ rack-level expertise with AMD’s silicon portfolio. This partnership aims to deliver a new class of end-to-end AI solutions and reduce the time required to design and deploy cluster-level data center AI systems. AMD later agreed to sell ZT Systems’ server manufacturing business to Sanmina for $3 billion, retaining ZT Systems’ rack-scale AI solutions design and customer enablement expertise, while positioning Sanmina as a preferred manufacturing partner.
Silo AI
AMD acquired Silo AI for $665 million in cash. Silo AI is the largest private AI lab in Europe and develops open-source multilingual large language models. This acquisition advances AMD’s strategy to deliver end-to-end AI solutions based on open standards. Silo AI’s team will contribute expertise in developing advanced AI models, including LLMs built on AMD platforms, and support the implementation of AI solutions for global customers.
Nod.ai
Acquired in October 2023, Nod.ai strengthens AMD’s open-source capabilities to better compete with Nvidia. Nod.ai develops compiler-based automation software that optimizes AI solutions for hyperscalers, enterprises, and startups. This accelerates deployment across AMD’s chip portfolio, including Instinct accelerators, EPYC processors, Versal system-on-chips, Ryzen AI processors, and Radeon GPUs, reducing the need for manual optimization and enabling faster deployment of high-performance AI models.
Mipsology
AMD acquired Mipsology, a French startup, in August 2023 to enhance its AI inference software capabilities. Mipsology’s plug-and-play software, Zebra, accelerates AI inference on FPGAs without requiring new tools or changes to neural network models. This acquisition supports the development of AMD’s complete AI software stack and expands its open ecosystem of software tools, libraries, and models.
Pensando
In May 2022, AMD acquired Pensando Systems, a networking chip designer, for $1.9 billion. Pensando’s distributed services platform expanded AMD’s data center offerings with a high-performance data processing unit (DPU) and software stack. Pensando’s products now support the networking side of AI infrastructure, including the 400-Gbps Pensando Salina DPU for AI front-end network clusters and the Pensando Pollara, an “Ultra Ethernet Consortium-ready AI NIC” essential for accelerator-to-accelerator communication in back-end networks.
Xilinx
AMD’s largest acquisition, a $49 billion deal completed in February 2022, brought programmable chip designer Xilinx into its portfolio. This expanded AMD’s presence in data centers, embedded computing, and telecommunications through the addition of Xilinx’s FPGAs.
Xilinx has developed an AI engine that is now incorporated into AMD products as a neural processing unit (NPU), most recently in the AMD Ryzen AI 300 series, with plans to integrate it into future EPYC CPUs.
Xilinx broadens AMD’s AI market coverage by offering Zynq adaptive chips for industrial applications, Versal adaptive chips for telecommunications, and Alveo accelerator chips and Kintex FPGAs for cloud data centers.
Related: What Makes an AI Company an Attractive Acquisition Target?
Final Thoughts
As AMD integrates these acquisitions, the industry is watching closely. AMD continues its rapid product release schedule for Instinct GPUs. AMD’s acquisition strategy positions the company as a strong contender in the AI landscape, with a focus on open-source software and end-to-end AI solutions. The technology sector will be monitoring how AMD leverages these initiatives to redefine AI performance standards.
Disclosure: This Page may contain affiliate links, for which we may receive compensation if you click on these links and make a purchase. However, this does not impact our content. We provide valuable and unbiased information.
+ There are no comments
Add yours