Localized AI infrastructures debut with greater capacity, higher security
Tech firm H3C releases servers and models optimized for generative AI applications like ChatGPT on Friday, which can soon be adopted by Chinese clients, such as telecom carriers and governments to build AI-powered and digitalized cities.
It's also H3C's first major product release, following its shareholder Unisplendour's recent announcement to buy all remaining shares from Hewlett Packard Enterprise or HPE for US$3.5 billion.
Compared with generative AI tools offered by Baidu and SenseTime, H3C's products are IT infrastructures suited for all those applications, with optimized AI computing capacities and data security setups to meet big company requirements.
H3C has taken an "AI for All" strategy, and regards AI as a top opportunity in the next decade, Tony Yu, chairman of Unisplendour, chief executive and president of H3C, said at the NAVIGATE conference held in Hangzhou.
The new products include an AI server to triple AI computing capacities and cut 70 percent AI training time, an AI computing algorithm scheduling platform, and a LinSeer model on private network for data security.
Hangzhou-based H3C has a wide range of products, covering sever, storage, switches and software, which meet customized AI demands in a safe way. Its partners Baidu and Intel also attended the Hangzhou conference.
A key task for H3C is to help clients improve AI efficiency and ensure security, said Steven You, Co-President and CTO of H3C.
In China, government and big companies focus on both AI capacities and data security. Localized products can dispel security concerns, thus promoting the development of domestic AI development, industry insiders said.
H3C, currently a joint venture between Beijing-based Unisplendour and HPE, will become 100 percent controlled by the Chinese owner. It has announced in May to acquire a 49 percent stake from US partner HPE for 24.7 billion yuan (US$3.5 billion), according to the Shenzhen-listed Unisplendour.