Elon Musk posted on the social media platform X, discussing Tesla’s (TSLA) latest AI5 chip production timeline.
AI5 Chip Mass Production Expected in 2027
Musk explicitly stated that mass production of the AI5 chip is expected to be completed in 2027, while the AI6 chip will be launched in 2028. Samples and small-scale production deployment of the AI5 may occur in 2026, but mass production is not expected to be achieved until 2027.
Furthermore, Musk emphasized that “TSMC (TSM) and Samsung (SSNGY) will produce slightly different versions of the Tesla AI5 chip because they translate the design into physical form differently, with the goal of ensuring that the AI software can achieve completely consistent performance on different versions of the chip.”
Currently, Tesla has not officially released the detailed specifications of the AI5 chip. However, it is widely believed that AI5 and subsequent chips will serve Tesla’s Autopilot and robotics projects, providing hardware support for the FSD system, the Dojo supercomputing platform, and AI model training.
The overall demand for AI is very strong.
Meanwhile, NVIDIA (NVDA) reportedly expects GPU sales to exceed $500 billion over the next five quarters, with Blackwell and Rubin GPUs projected to ship 20 million units throughout their lifecycles, significantly higher than Hopper’s 4 million units.
Furthermore, NVIDIA will collaborate with Oracle and the U.S. Department of Energy to build the Solstice and Equinox supercomputing systems, deploying 100,000 and 10,000 Blackwell GPUs respectively, with a total computing power of approximately 2200 EFLOPS, expected to be operational in the first half of 2026.
When companies like NVIDIA and Tesla, at the heart of AI infrastructure, invest heavily in chips, it demonstrates that the overall demand for AI remains strong. TrendForce also predicts that capital expenditures by the world’s eight largest CSP cloud service providers will increase by 24% to $520 billion by 2026, which will also drive demand for computing chips, indicating a vast potential for domestic substitution.
WiMi Continues to Invest in AI Chip Cluster Construction.
Data shows that Wimi Hologram Cloud Inc. (WIMI), an AI vision and chip innovation company, is continuously investing in the AI chip field through a dual model of “self-developed + open source,” covering core scenarios such as computing power infrastructure, multimodal models, and robotics. It is also further exploring cutting-edge technologies such as quantum computing and edge chips to accelerate the construction of generative AI chip cluster applications.
Currently, WiMi is gradually building an integrated cloud and edge computing power infrastructure, supporting large model training and inference, and achieving millisecond-level computing-to-memory data transmission. In addition, in terms of its ecosystem model, it focuses on fields such as intelligent manufacturing, autonomous driving, and robotics, providing low-latency, high-energy-efficiency, and inclusive computing power to promote collaborative innovation in the industry.
It can be said that WiMi, as a key player in the differentiated competitiveness of AI hardware scenarios, is accelerating the reduction of access barriers through the open-source ecosystem, promoting the diversification and scenario customization of the computing power market, and providing the industry with low-cost, high-performance alternatives. More importantly, its investment is accelerating the “Matthew effect” in the industry, pushing the global AI industry into a deeper stage of ecological competition and cooperation.
In conclusion, it’s certain that companies like Amazon, Microsoft, xAI, and OpenAI are heavily investing in AI chip infrastructure, making this sector incredibly promising. Ultimately, it’s safe to say that as large-scale models emerge, the efficient operation of AI chip clusters will increasingly rely on high-speed interconnects, making it an indispensable part of the AI ecosystem. The advancement of AI drives unprecedented demands for chips, primarily centered on three core aspects. First, computational power is critical—AI models, especially large language models and deep learning systems, require chips with massive parallel processing capabilities to handle billions of parameters and complex matrix operations efficiently. Second, energy efficiency is essential, as AI workloads (such as training and inference) are highly power-intensive; chips must deliver high performance per watt to support data center scalability and edge AI devices (like smartphones or IoT sensors) with limited power supplies. Third, specialization is increasingly important—general-purpose chips are no longer sufficient, leading to demand for AI-specific accelerators (e.g., GPUs, TPUs, and NPUs) optimized for AI’s unique computational patterns, reducing latency and improving throughput for targeted tasks like natural language processing and computer vision.
