SK Hynix expects AI memory market to grow 30% a year to 2030
The upbeat projection for global growth in high-bandwidth memory (HBM) for use in AI brushes off concern over rising price pressures in a sector that for decades has been treated like commodities such as oil or coal.
'AI demand from the end user is pretty much, very firm and strong,' said SK Hynix's Choi Joon Yong, the head of HBM business planning at SK Hynix.
The billions of dollars in AI capital spending that cloud computing companies such as Amazon, Microsoft and Alphabet's Google are projecting will likely be revised upwards in the future, which would be 'positive' for the HBM market, Choi said.
The relationship between AI build-outs and HBM purchases is 'very straightforward' and there is a correlation between the two, Choi said. SK Hynix's projections are conservative and include constraints such as available energy, he said.
But the memory business is undergoing a significant strategic change during this period as well. HBM - a type of dynamic random access memory or DRAM standard first produced in 2013 - involves stacking chips vertically to save space and reduce power consumption, helping to process the large volumes of data generated by complex AI applications.
BT in your inbox
Start and end each day with the latest news stories and analyses delivered straight to your inbox.
Sign Up
Sign Up
SK Hynix expects this market for custom HBM to grow to tens of billions of dollars by 2030, Choi said.
Due to technological changes in the way SK Hynix and rivals such as Micron Technology and Samsung Electronics build next-generation HBM4, their products include a customer-specific logic die, or 'base die', that helps manage the memory.
That means it is no longer possible to easily replace a rival's memory product with a nearly identical chip or product.
Part of SK Hynix's optimism for future HBM market growth includes the likelihood that customers will want even further customisation than what SK Hynix already does, Choi said.
At the moment it is mostly larger customers such as Nvidia that receive individual customisation, while smaller clients get a traditional one-size-fits-all approach.
'Each customer has different taste,' Choi said, adding that some want specific performance or power characteristics.
SK Hynix is currently the main HBM supplier to Nvidia, although Samsung and Micron supply it with smaller volumes.
Last week, Samsung cautioned during its earnings conference call that current generation HBM3E supply would likely outpace demand growth in the near term, a shift that could weigh on prices.
'We are confident to provide, to make the right competitive product to the customers,' Choi said.
100 per cent tariffs
US President Donald Trump on Wednesday said the United States would impose a tariff of about 100 per cent on semiconductor chips imported from countries not producing in America or planning to do so.
Choi declined to comment on the tariffs.
Trump told reporters in the Oval Office the new tariff rate would apply to 'all chips and semiconductors coming into the United States,' but would not apply to companies that were already manufacturing in the United States or had made a commitment to do so.
Trump's comments were not a formal tariff announcement, and the president offered no further specifics.
South Korea's top trade envoy Yeo Han Koo said on Thursday that Samsung Electronics and SK Hynix would not be subject to the 100 per cent tariffs on chips if they were implemented.
Samsung has invested in two chip fabrication plants in Austin and Taylor, Texas, and SK Hynix has announced plans to build an advanced chip packaging plant and an artificial intelligence research and development facility in Indiana.
South Korea's chip exports to the United States were valued at US$10.7 billion last year, accounting for 7.5 per cent of its total chip exports.
Some HBM chips are exported to Taiwan for packaging, accounting for 18 per cent of South Korea's chip exports in 2024, a 127 per cent increase from the previous year. REUTERS
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
CNA
9 hours ago
- CNA
CoreWeave revenue beats estimates on AI boom but shares fall on bigger loss
CoreWeave easily topped quarterly revenue estimates on Tuesday as the rapid adoption of artificial intelligence tools boosted demand for its cloud services, but a bigger-than-expected net loss sent its shares slumping 10 per cent after the bell. The company currently operates 33 AI data centers across the U.S. and Europe and offers access to backer Nvidia's chips, which are highly coveted by enterprises to train and run large AI models amid intense competition. CoreWeave reported revenue backlog of $30.1 billion as of end of June, compared with $25.9 billion on March 31. "Demand is humming, but it is the cost of growth that tempered the stock down in aftermarket trading," said Michael Ashley Schulman of Running Point Capital Advisors. Operating expenses jumped to $1.19 billion in the second quarter, from $317.7 million a year earlier. The company posted a net loss of $290.5 million, compared with analysts' average estimate of $190.6 million, according to data compiled by LSEG. "We are scaling rapidly as we look to meet the unprecedented demand for AI," CEO Michael Intrator said in the earnings statement. Investors have also focused on the company's reliance on a few big customers. "The backlog surge to $30B+ suggests demand visibility well beyond 2025, but the concentration in mega-customers like OpenAI means those relationships remain both the crown jewel and the single point of failure," said eMarketer analyst Jeremy Goldman. Meanwhile, CoreWeave executives reiterated the benefits of its $9 billion all-stock deal for crypto miner Core Scientific. The deal, announced in July, is facing opposition. Core Scientific's largest shareholder, Two Seas Capital, has said it would vote against the sale. CoreWeave reported second-quarter revenue of $1.21 billion, beating estimates of $1.08 billion. The Livingston, New Jersey-based company raised its annual revenue forecast to be between $5.15 billion and $5.35 billion. It had previously projected annual revenue of $4.9 billion to $5.1 billion.
Business Times
9 hours ago
- Business Times
Might a US-China mini thaw be underway?
ONE of the key elements of Donald Trump's foreign policy has been concern about China's rising power. Yet, the US president on Monday (Aug 11) extended his tariff truce with Beijing for another 90 days, raising the possibility that the two powers might potentially agree to a wider grand economic bargain in 2025 or beyond. The executive order was signed just hours before the previous tariff pause was set to expire, providing relief to financial markets around the world. Trump had previously increased tariffs on Chinese imports to 145 per cent, with Beijing reciprocating. An initial 90-day truce in May saw the US lower its levies to 30 per cent while Beijing reduced tariffs on US goods to 10 per cent and agreed to ramp up rare earth exports. The news comes soon after chip giants Nvidia and AMD agreed to pay the US government 15 per cent of Chinese revenues. This is part of what has been called an ' unprecedented ' deal to secure export licences to China. The US had earlier banned the sale of powerful chips used in areas like AI to China under export controls usually related to national security. Some security experts, including a number from Trump's first term, recently expressed deep concern that Nvidia's H20 chip was 'a potent accelerator' of China's AI capabilities. Window of opportunity The new 90-day pause allows time for a potential Trump meeting with President Xi Jinping at the Asia-Pacific Economic Cooperation summit in South Korea this October. Depending on the mood music this autumn, it is even possible that the US president may offer a visit to China at much the same time. This provides a window of opportunity for US and Chinese officials to finalise not only a tariff deal, but possibly the framework for a wider economic bargain that could include issues such as fentanyl trafficking and Chinese purchases of sanctioned Russian and Iranian oil. BT in your inbox Start and end each day with the latest news stories and analyses delivered straight to your inbox. Sign Up Sign Up A key question, therefore, is whether this truce presages a mini thaw in ties, as part of an effort to responsibly manage the competition between the two powers. While it is too early to tell, the answer will be decided largely by the personal chemistry of Trump and Xi. While economic and security fundamentals will mostly determine the long-term course of relations, both leaders are politically dominant in their nations and likely to remain in office until 2029 at least. With the two sides far apart on many issues, a Xi-Trump meeting would offer an opportunity for each leader to gauge the other's intent. The importance of this personal factor was shown during Trump's first term when his erratic nature accentuated the volatility in US-China ties. During Barack Obama's presidency, relations remained generally cordial, reflecting the commitment of Xi and Obama to bilateral stability. Both leaders appeared to recognise the super-priority of the relationship. Washington pursued a strategy that promoted cooperation on shared interests such as climate change, while constructively engaging on harder issues such as South China Sea tensions. It is sometimes forgotten that a key precursor for the Paris climate deal in 2015 was a US-China agreement in this area brokered by Obama and Xi. Renewing bilateral relations However, this engagement approach is perceived by Trump as having failed to shape Beijing's long-term behaviour. If and when Trump and Xi meet this autumn, the goal would be to find a framework to renew bilateral relations in the second half of the 2020s. Both appear to know that a stable relationship could have a broad, positive effect on global affairs, and potentially forestall further disruptive tensions. Yet, significant hazards in the US-China bilateral landscape remain. The future of Taiwan is a major point of uncertainty, and a tough framework of US measures against China carries over from the Trump and Joe Biden eras. These include sanctions over human rights abuses in Xinjiang and Hong Kong, and the rejection of nearly all of Beijing's maritime claims in the South China Sea. Ultimately, the new pause in the tariff war indicates that US-China relations need not inevitably be a force for greater global tension. Instead, it might yet show there is still a small, precious window of opportunity to evolve a strategic partnership to help underpin bilateral relations in the second half of the decade. The writer is an associate at LSE IDEAS at the London School of Economics
Business Times
10 hours ago
- Business Times
China urges firms to avoid Nvidia H20 chips after Trump resumes sales
[HONG KONG] China has urged local companies to avoid using Nvidia's H20 processors, particularly for government-related purposes, complicating the chipmaker's attempts to recoup billions in lost China revenue as well as the Trump administration's unprecedented push to turn those sales into a US government windfall. Over the past few weeks, Chinese authorities have sent notices to a range of firms discouraging use of the less-advanced semiconductors, sources familiar with the matter said, asking not to be named discussing sensitive information. The guidance was particularly strong against the use of H20s for any government or national security-related work by state enterprises or private companies, the sources said. In addition to Nvidia, Beijing's overall push affects artificial intelligence (AI) accelerators from Advanced Micro Devices (AMD), one of the sources said, though it's unclear whether any letters specifically mentioned AMD's MI308 chip. Both companies recently secured Washington's approval to resume lower-end AI chip sales to China, on the controversial and legally questionable condition that they give the US government a 15 per cent cut of the related revenue. Now, Nvidia and AMD face the challenge that their Chinese customers are under Beijing's pressure not to make those purchases. Some of Beijing's letters to companies included a series of questions, according to one of the sources, such as why they buy Nvidia H20 chips over local alternatives, whether that's a necessary choice given domestic options, and whether they have found any security issues in the Nvidia hardware. The notices coincide with state media reports that cast doubt on the security and reliability of H20 processors. Chinese regulators have raised those concerns directly with Nvidia, which has repeatedly denied that its chips contain such vulnerabilities. Right now, China's most stringent chip guidance is limited to sensitive applications, a situation that bears similarities to the way Beijing restricted Tesla vehicles and Apple iPhones in certain institutions and locations over security concerns. China's government also at one point barred the use of Micron Technology chips in critical infrastructure. BT in your inbox Start and end each day with the latest news stories and analyses delivered straight to your inbox. Sign Up Sign Up Still, it's possible that Beijing may extend its heavier-handed Nvidia and AMD guidance to a wider range of settings, according to one source with direct knowledge of the deliberations, who said that those conversations are in early stages. AMD declined to comment, while Nvidia said that 'the H20 is not a military product or for government infrastructure'. China has ample supplies of domestic chips, Nvidia said, and 'will not and never has relied on American chips for government operations'. China's Ministry of Industry and Information Technology and the Cyberspace Administration of China did not respond to faxed requests for comment on this story, which is based on interviews with more than a half-dozen sources familiar with Beijing's policy discussions. The White House did not respond to a request for comment outside normal business hours. The Chinese government's posture could make it more difficult for Nvidia and AMD to sell their hardware into the world's largest market for semiconductors. It also raises questions about the Trump administration's explanation for why the US is allowing those exports mere months after effectively banning such sales. Multiple senior US officials have said that their policy reversal was part of a trade accord with China, but Beijing has publicly indicated that the resumed H20 shipments were not part of any bilateral deal. China's recent notices to companies suggest that the Asian country may not have desired such a concession from Washington in the first place. Beijing's concerns are twofold. For starters, Chinese officials are worried that Nvidia chips could have location-tracking and remote shutdown capabilities, a suggestion that Nvidia has vehemently denied. Still, Trump officials are actively exploring whether location-tracking could be used to help curtail suspected smuggling of restricted components into China, and lawmakers have introduced a bill that would require location verification for advanced AI chips. Second, Beijing is intensely focused on developing its domestic chip capabilities, and wants Chinese companies to shift away from Western chips in favour of local offerings. Officials have previously urged Chinese firms to choose domestic semiconductors over Nvidia H20 processors, Bloomberg reported last September, and have introduced energy efficiency standards that the H20 chip does not meet. But Beijing has stopped short of outright banning the hardware, which Nvidia designed specifically for Chinese customers to abide by years of US curbs on sales of advanced chips to the Asian country. The H20 chip has less computational power than Nvidia's top offerings, but its strong memory bandwidth is quite well suited to the inference stage of AI development, when models recognise patterns and draw conclusions. That's made it a desirable product to companies such as Alibaba Group Holding and Tencent Holdings in China, where domestic chip champion Huawei Technologies is struggling to produce enough advanced components to meet market demand. By one estimate from Biden officials, who considered but did not implement controls on H20 sales, losing access to that Nvidia chip would make it three to six times more expensive for Chinese companies to run inference on advanced AI models. 'Beijing appears to be using regulatory uncertainty to create a captive market sufficiently sized to absorb Huawei's supply, while still allowing purchases of H20s to meet actual demands,' said Lennart Heim, an AI-focused researcher at Rand, of China's push for companies to avoid American AI chips. 'This signals that domestic alternatives remain inadequate even as China pressures foreign suppliers.' US President Donald Trump on Monday called the H20 chip 'obsolete', saying that China 'already has it in a different form'. That echoed previous statements by officials in his administration, who defended the decision to resume H20 exports on the grounds that Huawei already offers comparable chips to the H20. The US should keep the Chinese AI ecosystem reliant on less-advanced American technology for as long as possible, these officials argue, in order to deprive Huawei of the revenue and know-how that would come from a broader customer base. Commerce Secretary Howard Lutnick and other Trump officials have also claimed that the H20 move was part of a deal to improve American access to Chinese rare-earth minerals, despite the Trump team's previous assertions that such an arrangement was not on the table. 'As the Chinese deliver their magnets, then the H20s will come off,' Lutnick said last month. Treasury Secretary Scott Bessent said in late July that the magnet issue had been 'solved'. The first Nvidia H20 and AMD MI308 licenses arrived a bit over a week after Bessent's declaration, after Nvidia chief executive officer Jensen Huang met with the president and both companies agreed to share their China revenue with the US government. BLOOMBERG
