Meta acquires Rivos to develop proprietary AI accelerators, joining the hyperscaler race for custom silicon.

Meta Platforms (NASDAQ: META) has acquired chip startup Rivos in a strategic move announced on September 30, 2025, to bolster its artificial intelligence capabilities and reduce reliance on external chip suppliers. Rivos, which had been seeking new funding at a $2 billion valuation in August, specializes in developing high-performance, energy-efficient processors and System-on-Chip solutions based on the open-source RISC-V instruction set architecture. The startup has been working on its own Data Parallel Accelerator—a GPU designed specifically for AI tasks—supported by RISC-V-based CPU cores. Though financial terms remain undisclosed, the acquisition represents Meta's fundamental shift toward vertical integration, allowing the company to tailor hardware specifically for its demanding AI workloads, from recommendation systems to advanced generative AI models.

The deal addresses Meta's longstanding reliance on external semiconductor manufacturers, most notably Nvidia (NASDAQ: NVDA), to which the company spends billions annually on high-performance GPUs. Meta's in-house team has already begun testing its first AI training chip, the Meta Training and Inference Accelerator (MTIA). Integration of Rivos's talent and technology will accelerate this roadmap, providing crucial intellectual property and engineering expertise to advance Meta's custom chip deployment for recommendation systems and generative AI products like Meta AI. This vertical integration effort comes as Meta faces significant infrastructure costs, with 2025 expenses projected at $114–$119 billion, including up to $65 billion in capital expenditure primarily for AI infrastructure. By controlling its hardware stack, Meta can achieve performance and efficiency gains unattainable with off-the-shelf solutions and reduce long-term costs.

The acquisition reshapes the competitive landscape for AI hardware, creating clear winners and losers. Meta itself gains the ability to custom-design chips perfectly suited to its unique AI models and data center environments. Taiwan Semiconductor Manufacturing Company (TSMC) (NYSE: TSM), Meta's primary manufacturing partner for custom chips, also benefits by solidifying its position as the foundry of choice for leading-edge silicon. Conversely, traditional AI chip powerhouses face significant long-term challenges. While Meta will likely continue purchasing Nvidia GPUs for the foreseeable future, the acquisition signals a clear intent to reduce that reliance over time. Other major chipmakers such as Advanced Micro Devices (AMD) (NASDAQ: AMD) and Intel (NASDAQ: INTC), competing for market share in AI accelerators, now face another formidable in-house solution from a hyperscale giant. Smaller AI chip startups may face increased pressure as well, as the growing trend of vertical integration by major tech players limits market opportunities and makes it harder to secure design wins against internally developed solutions.

Meta's acquisition of Rivos exemplifies broader industry trends sweeping the technology sector. Tech giants including Apple (NASDAQ: AAPL) with its M-series chips, Google (NASDAQ: GOOGL) with its Tensor Processing Units (TPUs), and Amazon (NASDAQ: AMZN) with its Graviton and Trainium/Inferentia chips have all demonstrated the strategic advantages of designing their own processors. These advantages include superior performance optimized for specific workloads, greater control over product roadmaps, and ultimately, reduced costs—benefits Meta now pursues with heightened intensity through its acquisition of Rivos and acceleration of its custom silicon roadmap.