Japanese Company Preferred Networks Intensifies Investment in Specialized AI Chip Technology
In Brief
Preferred Networks has ramped up its funding for the development of bespoke AI chips, with the goal of ensuring easier access to vital hardware amid a global spike in AI investments.
Japan-based startup Preferred Networks The firm is increasing its funding for specially made artificial intelligence (AI) chips, aiming to secure access to essential hardware amidst the global investment boom spurred by innovations in generative AI.
Preferred Networks focuses on chips tailored for diverse AI applications, achieving lower power consumption and improved computing capabilities partly by shifting tasks typically executed by hardware to software solutions.
The startup has completed the design phase of its second-generation AI chip, which is presently in production. TSMC , and set to power its upcoming supercomputer .
"We recognized the necessity to enhance energy efficiency and reduce procurement risks, which are critical for the sustainability of our business,\"
stated Toru Nishikawa, the co-founder and CEO of Preferred Networks.
The company, which has investors including the automotive manufacturer and robotics company Fanuc, began work on its first-generation AI chip back in 2016 to power its supercomputers. Toyota Preferred Networks anticipates making its latest technologies accessible for building large language models (LLMs) and drug discovery in the upcoming year, with intentions to provide pure computational power to clients by 2027.
Companies Are Exploring In-House AI Chip Production
With the rising demand for generative AI—which typically relies on GPUs for training and execution—there’s a noticeable shortage of GPUs. To decrease reliance on GPUs, well-capitalized corporations, especially tech behemoths, are developing custom chips
specifically designed for creating and implementing AI models, and often making these chips available to customers. chips are among those creating silicon chips internally.
Amazon and Microsoft launched two families of chips—AWS Graviton4 and AWS Trainium2—developed in-house for training foundational models and processing inference.
In a recent development, AWS Microsoft also debuted its custom AI chips,
100 and Cobalt 100, specifically crafted for its cloud infrastructure. These chips are designed to power Microsoft’s Azure data centers and are expected to be released in 2024. Azure Maia The growing interest in customized AI chips by firms across the globe represents a calculated response to the escalating demand for technological hardware.
Please remember that the information showcased on this site is not meant to serve as legal, tax, investment, financial, or any other advisory service. It’s crucial only to invest what you can afford to lose and to seek independent financial advice if you have any uncertainties. For more insights, we recommend checking the terms and conditions as well as the help pages provided by the issuer or advertiser. MetaversePost is dedicated to providing accurate and unbiased reporting, but market conditions may fluctuate unexpectedly.
by
Posts by Alisa Davidson Alisa Davidson Alisa, a committed journalist at Cryptocurrencylistings, specializes in areas like cryptocurrency, zero-knowledge proofs, investments, and the vast domain of Web3. With a sharp insight into emerging trends and technologies, she offers comprehensive coverage to engage readers in the fast-paced world of digital finance.