Asustek Computer Inc (華碩電腦) plans to introduce one of the first services that lets companies tap into the potential of generative artificial intelligence (AI) while keeping control over their data.
The novelty of the Taipei-based firm’s offering, called AFS Appliance, is that all of the hardware would be installed in the client’s own facilities — to maintain security and control.
The AI computational platform, built on Nvidia’s chip technology, would be operated and updated with new data by Asustek.
Photo: Sam Yeh, AFP
A major concern around services such as OpenAI is that they are operated through online data centers that can expose sensitive information. Samsung Electronics Co banned employees from using OpenAI’s ChatGPT after it found workers had uploaded sensitive code to the platform.
Asustek is to preload the AFS hardware with its own large language model, called Formosa, which it says is equivalent to ChatGPT 3.5 and trained in traditional Chinese.
The company aims to price the service at about US$6,000 per month, Asus Cloud Corp (華碩雲端) and Taiwan Web Service Corp (台智雲) president Peter Wu (吳漢章) said in an interview on Monday.
The highest-spec offering, which adds an Nvidia DGX AI supercomputer platform, would cost about US$10,000 a month, he said.
Wu aims to have 30 to 50 enterprise customers in Taiwan and expand internationally by the end of this year.
“Nvidia is a partner with us to accelerate the enterprise adoption of this technology,” Wu said. “Before ChatGPT, the enterprises were not aware of why they need so much computing power.”
Most businesses would deploy generative AI on their own premises due to sensitivity about access to their proprietary data, Wu said.
“They need a smart brain to be under their own control and management, as it will touch on the most sensitive data,” he said.
Recently returning from a trip to Singapore, Wu says he sees a lot of interest from banks and hospitals. In the clinical context, generative AI would help doctors document treatments and patients visits quicker, while also helping communicate diagnoses in more relatable language to patients.
There is no supply shortage of Nvidia’s A100 chips, which run the AFS Appliance, Wu said, and the existing partnership between the companies is helping ensure that Asustek would have all the firepower it needs to roll out its new offering.
GROWING OWINGS: While Luxembourg and China swapped the top three spots, the US continued to be the largest exposure for Taiwan for the 41st consecutive quarter The US remained the largest debtor nation to Taiwan’s banking sector for the 41st consecutive quarter at the end of September, after local banks’ exposure to the US market rose more than 2 percent from three months earlier, the central bank said. Exposure to the US increased to US$198.896 billion, up US$4.026 billion, or 2.07 percent, from US$194.87 billion in the previous quarter, data released by the central bank showed on Friday. Of the increase, about US$1.4 billion came from banks’ investments in securitized products and interbank loans in the US, while another US$2.6 billion stemmed from trust assets, including mutual funds,
Micron Memory Taiwan Co (台灣美光), a subsidiary of US memorychip maker Micron Technology Inc, has been granted a NT$4.7 billion (US$149.5 million) subsidy under the Ministry of Economic Affairs A+ Corporate Innovation and R&D Enhancement program, the ministry said yesterday. The US memorychip maker’s program aims to back the development of high-performance and high-bandwidth memory chips with a total budget of NT$11.75 billion, the ministry said. Aside from the government funding, Micron is to inject the remaining investment of NT$7.06 billion as the company applied to participate the government’s Global Innovation Partnership Program to deepen technology cooperation, a ministry official told the
Taiwan Semiconductor Manufacturing Co (TSMC, 台積電), the world’s leading advanced chipmaker, officially began volume production of its 2-nanometer chips in the fourth quarter of this year, according to a recent update on the company’s Web site. The low-key announcement confirms that TSMC, the go-to chipmaker for artificial intelligence (AI) hardware providers Nvidia Corp and iPhone maker Apple Inc, met its original roadmap for the next-generation technology. Production is currently centered at Fab 22 in Kaohsiung, utilizing the company’s first-generation nanosheet transistor technology. The new architecture achieves “full-node strides in performance and power consumption,” TSMC said. The company described the 2nm process as
JOINT EFFORTS: MediaTek would partner with Denso to develop custom chips to support the car-part specialist company’s driver-assist systems in an expanding market MediaTek Inc (聯發科), the world’s largest mobile phone chip designer, yesterday said it is working closely with Japan’s Denso Corp to build a custom automotive system-on-chip (SoC) solution tailored for advanced driver-assistance systems and cockpit systems, adding another customer to its new application-specific IC (ASIC) business. This effort merges Denso’s automotive-grade safety expertise and deep vehicle integration with MediaTek’s technologies cultivated through the development of Media- Tek’s Dimensity AX, leveraging efficient, high-performance SoCs and artificial intelligence (AI) capabilities to offer a scalable, production-ready platform for next-generation driver assistance, the company said in a statement yesterday. “Through this collaboration, we are bringing two
RSS