Memorychip supplier AP Memory Technology Corp (愛普科技) yesterday said it expects to see the first revenue contribution from licensing its intellectual property (IP) linked to advanced packaging technologies used in artificial-intelligence (AI) accelerator chips in the next year.
That would be major progress in its strategy of shifting its focus to supplying very-high-bandwidth memory (VHM) chips used in mainstream AI applications from chips for crypto mining.
Licensing IP used in AI chips appears to be the first step. AP Memory said it is in discussion with several customers to grant its IP to develop silicon interposer, a key part of the advanced packaging technology for AI chips.
Photo courtesy of AP Memory Technology Corp
The company yesterday declined to identify Samsung Electronics Co as one of the potential customers who has adopted or intend to adopt its technology.
“We cannot comment on specific customers,” AP Memory vice president Liu Chin-hung (劉景宏) told investors. “We are in talks with more than one customer to license our IP and provide other services as well. We expect to see revenue contribution within a year at the earliest.”
“We have received a license fee from some customers,” Liu said.
AP Memory shares closed at NT$348 on Wednesday, up from NT$174 at the beginning of this year. Investors consider AP Memory one of the beneficiaries of the AI boom.
Apart from IP licensing, AP Memory also aims to offer its VHM technology for AI chips, replacing the mainstay high-bandwidth memory (HBM) developed by the world’s top three memorychip makers — Samsung, Micron Technology Inc and SK Hynix Inc.
Some AI accelerator chip suppliers have approached AP Memory seeking potential supply of VHM chips as supplies of HBM chips are scarce, Liu said.
Spiking demand for AI algorithms, a large-language-model to generate data and text, has boosted demand for HBM chips, Liu said.
Some projects have entered the proof-of-concept phase, the first of six or more phases of development before a product enters mass production, Liu said.
It would take one to three years for one such project to yield results, AP Memory chairman and chief executive officer Chen Wen-liang (陳文良) said.
As a result, the company’s AI business group would be in the trough in next two quarters in terms of revenue, Chen said.
Falling demand for older-generation chips used in crypto mining is also a factor, he said.
New-generation chips are to enter production next year, he said.
The company’s AI business group accounted for 10 percent of its total revenue last quarter, with 90 percent from supplying memory chips used in Internet-of-Things devices.
AP Memory’s net profit in the second quarter soared 277 percent to NT$389.74 million (US$12.32 million) from NT$101.04 million in the first quarter. On an annual basis, net profit contracted 17 percent from NT$456.57 million.
Earnings per share rose to NT$2.35 from NT$0.62 the previous quarter. That represented a decline from NT$2.83 a year earlier.
The company expects revenue to grow in the second half of the year thanks to inventory replenishment demand from customers.
For the full year, the company expects revenue to be flat or grow slightly from NT$5.1 billion last year.
Nvidia Corp’s demand for advanced packaging from Taiwan Semiconductor Manufacturing Co (TSMC, 台積電) remains strong though the kind of technology it needs is changing, Nvidia CEO Jensen Huang (黃仁勳) said yesterday, after he was asked whether the company was cutting orders. Nvidia’s most advanced artificial intelligence (AI) chip, Blackwell, consists of multiple chips glued together using a complex chip-on-wafer-on-substrate (CoWoS) advanced packaging technology offered by TSMC, Nvidia’s main contract chipmaker. “As we move into Blackwell, we will use largely CoWoS-L. Of course, we’re still manufacturing Hopper, and Hopper will use CowoS-S. We will also transition the CoWoS-S capacity to CoWos-L,” Huang said
Nvidia Corp CEO Jensen Huang (黃仁勳) is expected to miss the inauguration of US president-elect Donald Trump on Monday, bucking a trend among high-profile US technology leaders. Huang is visiting East Asia this week, as he typically does around the time of the Lunar New Year, a person familiar with the situation said. He has never previously attended a US presidential inauguration, said the person, who asked not to be identified, because the plans have not been announced. That makes Nvidia an exception among the most valuable technology companies, most of which are sending cofounders or CEOs to the event. That includes
TARIFF TRADE-OFF: Machinery exports to China dropped after Beijing ended its tariff reductions in June, while potential new tariffs fueled ‘front-loaded’ orders to the US The nation’s machinery exports to the US amounted to US$7.19 billion last year, surpassing the US$6.86 billion to China to become the largest export destination for the local machinery industry, the Taiwan Association of Machinery Industry (TAMI, 台灣機械公會) said in a report on Jan. 10. It came as some manufacturers brought forward or “front-loaded” US-bound shipments as required by customers ahead of potential tariffs imposed by the new US administration, the association said. During his campaign, US president-elect Donald Trump threatened tariffs of as high as 60 percent on Chinese goods and 10 percent to 20 percent on imports from other countries.
INDUSTRY LEADER: TSMC aims to continue outperforming the industry’s growth and makes 2025 another strong growth year, chairman and CEO C.C. Wei says Taiwan Semiconductor Manufacturing Co (TSMC, 台積電), a major chip supplier to Nvidia Corp and Apple Inc, yesterday said it aims to grow revenue by about 25 percent this year, driven by robust demand for artificial intelligence (AI) chips. That means TSMC would continue to outpace the foundry industry’s 10 percent annual growth this year based on the chipmaker’s estimate. The chipmaker expects revenue from AI-related chips to double this year, extending a three-fold increase last year. The growth would quicken over the next five years at a compound annual growth rate of 45 percent, fueled by strong demand for the high-performance computing
RSS