(Image from Google)
AsianFin -- British chip design firm Imagination Technologies on Wednesday launched its next-generation E-Series GPU IP, a major upgrade aimed at powering the future of edge AI and graphics applications.
The new product delivers a dramatic performance leap—up to 400% AI acceleration improvement over its predecessor—positioning the company to ride the global AI investment wave and capture new opportunities in consumer electronics, automotive, and mobile computing.
Built on an efficient parallel processing architecture, the E-Series supports INT8 precision and scalable AI performance ranging from 2 TOPS to 200 TOPS. This makes it well-suited for compute-heavy tasks such as natural language processing on smartphones, real-time rendering in desktop systems, and industrial vision and assisted driving in edge devices.
Bai Nong, Chairman of Imagination China and President of Asia-Pacific, said the new GPU represents a milestone in the company's evolution beyond graphics into general-purpose computing. "It's fully optimized for performance, power, and area efficiency," Bai told AsianFin.
"As lightweight large language models like DeepSeek continue to evolve, edge-side AI demands are accelerating. This GPU IP solution is designed to meet that moment."
The announcement comes as AI spending soars globally. According to IDC, total global AI IT investment is projected to hit $315.8 billion in 2024, growing to $815.9 billion by 2028 at a CAGR of 32.9%. China alone is expected to surpass $100 billion in AI investment by 2028, with generative AI accounting for more than $30 billion of that, growing at a staggering 51.5% annually.
Imagination's E-Series GPUs integrate ray tracing capabilities alongside new "Neural Cores" that provide dense AI compute without relying on a fixed CPU architecture. The cores feature double the register storage of competing IPs and include proprietary Burst Processor tech, which boosts edge AI power efficiency by 35%.
Vice President of Product Management Dennis Laudick noted that the GPU IP can be integrated into SoC designs across both ARM and RISC-V architectures. "We've validated its compatibility across a wide range of chiplet designs, with no decoupling issues," he said.
The first commercial E-Series products are set to ship in fall 2025, with versions tailored for automotive, desktop, mobile, and consumer electronics already under development. Licensing for the technology is already underway.
Founded in 1985 and headquartered in the UK, Imagination Technologies is one of the world's top four chip IP companies. It holds over 37% of the mobile GPU IP market and leads in automotive GPU IP with 45% share. Its client base spans more than 2,000 global partners.
In fiscal year 2023, the company reported revenue of £124 million ($155 million), up 3.6% year-on-year, with licensing income rising by £4.1 million and pre-tax profit hitting £13.5 million.
Bai reiterated China's strategic importance to Imagination, emphasizing the company's long-term commitment despite global market turbulence. "We're doubling down on China with greater investment, expanded local teams, and deeper ecosystem cooperation," he said. "We are confident in the Chinese market and look forward to seeing the E-Series thrive here."
The launch signals Imagination's ambition to remain a critical player in the evolving semiconductor landscape as AI reshapes everything from chip design to end-user devices.
更多精彩内容，关注钛媒体微信号 （ID：taimeiti），或者下载钛媒体 App
