Asian Suppliers Surge to 90% of Nvidia's Production Costs as Physical AI Drives New Rally

Nvidia's Push Into Physical AI Sparks Rally in Asian Partners

Nvidia's push into physical AI -- spanning robotics, autonomous systems and AI-enabled manufacturing -- extends its influence beyond chips into real-world deployment, positioning Asia as a critical partner in that expansion. The list of Asian stocks that benefit from business partnership with Nvidia Corp. is getting longer, as the region further integrates into the AI chip giant's business ecosystem. Just in the past week, South Korea's LG Electronics Inc., Taiwan's Nanya Technology Corp., as well as China's Huizhou Desay SV Automotive Co.Bloomberg Terminal and Pateo Connect Technology Shanghai Corp. have become the latest to rally on news of tie-ups, supply-chain participation or product collaboration with the US chip designer. Investor enthusiasm about these firms, some of which are relatively obscure outside the industry or domestic markets, is a reminder of how Nvidia-induced demand is shaping stock performance across Asia's technology supply chain. Asian suppliers now account for about 90% of Nvidia's production costs, up from roughly 65% last year, according to data compiled by Bloomberg. The explosive growth of the chip designer's products has intensified its reliance on Asian partners that dominate manufacturing, assembly and key components. "It's inevitable that global tech companies like Nvidia will continue to ramp up their reliance on Asia supply chain," said Vey-Sern Ling, managing director at Union Bancaire Privee. Physical AI "can add on top of the already burgeoning demand from Asia's supply chains for AI chips," he added. Nvidia has expanded its roster of Asian partners in recent years, primarily through deeper chip-focused ties with suppliers such as SK Hynix Inc. and Samsung Electronics Co. While those partnerships concentrated on scaling AI computing power, the latest wave of collaborations in the region pointed to a shift beyond semiconductors into physical AI, including robotics. Shares of LG Electronics jumpedBloomberg Terminal as much as 15% on Tuesday, their biggest intraday gain since Feb. 11, following a domestic media report that the firm and Nvidia will discuss a plan to integrate its home robot with the US chip designer's platform. In Taiwan, Nanya Technology's shares surgedBloomberg Terminal 10% after a local news report on the chipmaker's collaboration with Nvidia. Elsewhere, China's Huizhou Desay also saw its stock rally after unveilingBloomberg Terminal a new mass‑production intelligent driving solution with Nvidia, while automobile product maker Pateo Connect Technology's shares soaredBloomberg Terminal after the company entered a series of collaborations with Nvidia. A LG Electronics spokesperson said the company can confirm that it recently met with Nvidia, adding that the two companies are exploring strategic collaboration in physical AI, including the robotics ecosystem. A representative for Samsung Electronics declined to comment in response to Bloomberg's queries. Nanya said in an emailed response that it doesn't comment on customer-related information. Nvidia's push into physical AI -- spanning robotics, autonomous systems and AI-enabled manufacturing -- extends its influence beyond chips into real-world deployment, positioning Asia as a critical partner in that expansion. Its Chief Executive Officer Jensen Huang has framed physical AI as the next wave after generative AI. "Increasing and broadening demand is creating opportunities across industries for more tech suppliers to join the supply chain as AI buildout continues globally," said Marvin Chen, a strategist at Bloomberg Intelligence. "It means that tech heavy north Asian markets may continue to outperform." The latest capital expenditure guidance from US tech giants shows AI spending is accelerating, with Amazon.com Inc., Microsoft Corp. and Alphabet Inc. each committing roughly $190 billion to $200 billion for this year and Meta Platforms Inc. raising its outlays to as much as $145 billion. Nvidia accounts for about half of Microsoft's capital expenditure and roughly a quarter of Amazon's, with a smaller but still leading share at Meta and Alphabet, based on calculations of data compiled by Bloomberg. Meanwhile, Hon Hai is a consistent secondary beneficiary, particularly at Microsoft and Amazon, while SK Hynix takes a mid-single-digit share across companies. Surging demand has shown up in the results of those suppliers. Samsung's semiconductor arm beat expectations last week with a 48-fold jump in profit. A few days earlier, SK Hynix reported a five-fold increase in quarterly earnings. "Asia's technology base is a structurally important advantage, particularly as AI creates new demand across semiconductors, components, servers and broader hardware infrastructure," said Rajeev De Mello, a portfolio manager at Gama Asset Management SA. "Asia has already developed significant experience and supply chains to build advanced semiconductors, and robots, which is a strong base for implementing physical AI." Read more about the rise of Korean and Taiwanese stock markets Korea Passes UK to Become World's Eighth-Largest Stock Market AI Boom Drowns Out War Fears to Fuel Asia's Great Market Divide AI Chip Surge Elevates Taiwan, Korea in Global Equity Rankings

AI frenzy signals end of boom and bust for memory chipmakers

Investors are betting on a prolonged boom for memory chipmakers amid voracious AI demand that has prompted customers to lock in multiyear contracts with SK Hynix and Samsung Electronics. Hynix, the world's second-largest memory chipmaker and a key supplier to Nvidia, said the "structural shift" differs from past booms because customers are prioritising security of supplies over price amid an acute shortage. That shift is reinforcing expectations that an industry long defined by boom-and-bust cycles may be shaking off its volatile past. Analysts say the memory sector is facing a supply crunch as chipmakers, which also include Micron Technology, struggle to meet demand driven by a global data centre build-out, making memory a key bottleneck in the AI supply chain. Many view the shortage as structural rather than temporary because demand has shifted from cyclical consumer electronics to deep-pocketed AI hyperscalers rapidly increasing capital spending. Demand is also becoming more persistent as AI spreads into everyday applications, increasing the need for memory across devices and services. Samsung's co-chief executive Young Hyun Jun told shareholders in March that AI demand was driving an "unprecedented supercycle" in the semiconductor market and the company is planning a "substantial" increase in capital expenditure this year to meet surging memory demand. SK Hynix and Samsung, which reports quarterly results on Thursday, say customers are now seeking contracts for three to five years rather than the usual quarterly agreements. This move shift enhances demand visibility and reduces the risk of oversupply, enabling producers to increase capital spending in a more controlled way. "We are in a supercycle as AI requires more powerful memory for inference and agentic services," said Kwon Seok-joon, a professor at Sungkyunkwan University. AI inference is the process whereby a trained AI model such as a chatbot generates responses. Agentic AI services are advanced systems that can handle complex problems and act on their own, with little human assistance. "Hyperscalers are expanding capacity aggressively to avoid falling behind in the AI race, pushing up chip prices while production remains constrained." Samsung's Young Hyun Jun told shareholders in March that the company was talking to major customers about moving on to contracts of between three and five years. The company also plans to increase spending this year, earmarking a record Won110tn for chip capacity expansion and research. Analysts say the supply crunch is unlikely to ease before 2028, given the time required to build new plants. SK Hynix said it would "significantly" increase capital spending this year, but technological challenges are expected to limit how quickly supply can expand. "We are in the third year of the upcycle with no end in sight," said Daniel Kim, an analyst at Macquarie. "Chipmaking has reached a level of complexity where supply cannot easily be increased. Shortages are likely to intensify next year." As a result, the industry has become a seller's market, with tight supply giving producers greater pricing power. Long-term contracts also allow companies to better manage output, unlike in the past when they were forced to react to volatile demand. "The tables have turned," said Kwon, noting that Samsung, SK Hynix and Micron jointly control about 90 per cent of the market for DRam, which serves as the main memory for computers and smartphones. "Producers now have the upper hand." Analysts believe this oligopoly, formed after decades of consolidation, will allow memory makers to sustain elevated margins as long as investment in AI infrastructure continues. SK Hynix reported a quarterly operating margin of 72 per cent, while Samsung's memory margin is estimated to be well above 60 per cent, compared with Nvidia's 65 per cent and TSMC's 58 per cent. Nomura analyst CW Chung said memory chip prices could rise by up to half in the second quarter compared with the first three months of the year, potentially lifting margins to above 80 per cent. He added that the growing share of long-term contracts would reduce sharp price swings. "Memory chips have become a core pillar of AI infrastructure," Chung said, predicting that the upcycle could last three to five years. Net profits at Samsung and SK Hynix this year are forecast at $151bn and $115bn respectively -- far bigger than TSMC's $81bn, according to Bloomberg. Yet both Korean companies are valued at less than six times projected earnings, compared with roughly 19 times for TSMC and 22 times for Nvidia. The gaps reflect the industry's historical cyclicality, although analysts expect them to narrow as earnings stability improves. SK Hynix shares have surged more than 600 per cent over the past year, while Samsung shares have gained nearly 300 per cent. Risks for the sector remain, however. Rapid capacity expansion could eventually lead to oversupply, particularly if hyperscalers slow their investment in AI infrastructure. Growing competition from Chinese players such as CXMT and YMTC also raises concerns about excess long-term supply in lower-end products. "This is certainly a supercycle. But that doesn't mean there won't be ups and downs," said Chris Miller, author of Chip War: The Fight for the World's Most Critical Technology. "The data centre business will have its own cycles, even if longer-term contracts smooth some volatility." Kwon expects demand to remain strong, arguing that on-device AI could eventually surpass data centres as the main driver of memory consumption. But he warned that oversupply in legacy chips could emerge about 2035 as new plants come online and Chinese capacity expands. Kwon estimates Samsung and SK Hynix could more than double their combined capacity by then: "There is strong optimism about the durability of this cycle -- but it ultimately depends on how long the current pace of AI data centre expansion can be sustained." Miller also warned that Chinese producers are using the current boom as an opportunity to strengthen their position, as tight supply pushes some western manufacturers to consider broader use of Chinese chips in consumer electronics. Citi analyst Peter Lee predicted that the upcycle could extend beyond seven years if AI momentum holds but cautioned against assuming a permanent break from the past. "Many investors still don't believe the industry's cyclical nature has fundamentally changed," he said. "This cycle may be stronger and longer-lasting, but cycles won't disappear."

AI Rally Ripples Through Chip Supply Chain, Minting New Winners

The artificial intelligence rally exploding across Asia is now spreading deeper into the supply chain. For much of the past year, the spotlight has been on chipmakers such as Taiwan Semiconductor Manufacturing Co., Samsung Electronics Co. and SK Hynix Inc. -- all critical suppliers to Nvidia Corp. A global chip shortage has sent those stocks skyrocketing to record highs. Now, investors are looking further into the AI supply ecosystem as even the most powerful processors cannot work without the lesser-known components that support them. That growing awareness -- combined with rising demand and prices -- is helping a new group of companies rally. They fall into three main categories. Multi-layer ceramic capacitors (MLCCs) regulate power inside electronic systems. Advanced chip substrates connect semiconductors to the rest of the hardware. And thermal compression bonding (TCB) is the precision process that fuses everything together. "Think of the printed circuit-board as a dining table. Then the plate on the dining table, you can call it a substrate -- and the food inside the plate is the chips," said Kieron Poon, investment director of Asian equities at Aberdeen Investments. Within this group, substrate makers such as Unimicron Technology Corp. and Ibiden Co. have surged around 770% and 530% respectively over the past 12 months. MLCC producers Samsung Electro-Mechanics Co. and Murata Manufacturing Co. climbed to record highs this month. TCB leader Hanmi Semiconductor Co. also hit an all-time high. Much of this supply chain is concentrated in Asia, mainly South Korea, Taiwan, Japan and China. What's driving this is the intensity of AI infrastructure buildouts. AI servers consume far more power than conventional ones, and that has a cascading effect. More power means more components to manage and stabilize it. An AI server can use 10 to 15 times more MLCCs than a standard server, and about 30 times more than a smartphone, according to Young Jae Lee, senior investment manager at Pictet Asset Management. This surge in demand is tightening supply and pushing up prices for those components. Samsung Electro-Mechanics said this week it's considering raisingBloomberg Terminal MLCC product prices by as much as 10%. Citigroup Inc. analyst Takayuki Naito also highlighted growing expectations for price increases across components like MLCCs, aluminum capacitors and package substrates. Naito sees this as a potential support for Japanese manufacturers such as Murata Manufacturing and Taiyo Yuden Co., which may take a more aggressive approach to pricing. JPMorgan Chase & Co. analysts raised their price targets for Murata and Yuden last week, saying supply and demand are likely to stay tight for an extended period. Production lines for MLCCs and substrates are already running at more than 90% capacity, said Simon Woo, head of Korea research at BofA Global Research. "If AI demand ramps up even slightly from here, capacity for conventional use will shrink sharply," he said. The ripple effects are spreading further across the ecosystem. Optical component makers have also been pulled into the rally, as investors pay closer attention to the role these technologies play in data centers, especially as rising use of AI drives the need for ever more bandwidth. Supply of capacitors, substrates and TCB remains concentrated and the firms serve a rapidly expanding pool of customers, said Aberdeen's Poon. This means pricing power will "definitely still be in the hands of the supplier," he said.

Nvidia's pivot to physical AI ignites rally across Asian supply chain By Investing.com

Investing.com -- Nvidia Corp.'s strategic expansion into "physical AI" is driving a significant rally among its Asian partners as the region becomes increasingly integral to the chip giant's production ecosystem, according to a report from Bloomberg. Asian suppliers now account for approximately 90% of Nvidia's production costs, a sharp increase from roughly 65% just last year. The shift highlights a move beyond pure semiconductors into robotics, autonomous systems, and AI-enabled manufacturing, a transition CEO Jensen Huang has characterized as the next major wave of artificial intelligence. The market's reaction to the tech firm's expansion has been swift, with several regional players seeing double-digit gains following news of collaboration. LG Electronics Inc. shares surged as much as 15% on reports of integrating home robots with Nvidia's platform, while Taiwan's Nanya Technology Corp. climbed 10% on partnership news. In China, Huizhou Desay SV Automotive Co. and Pateo Connect Technology Shanghai Corp. also saw shares soar after unveiling intelligent driving solutions and collaborations with the U.S. chip designer. Hyperscaler spending sustains North Asian tech outperformance The momentum is supported by massive capital expenditure from U.S. "hyperscalers," with Amazon, Microsoft, and Alphabet committing nearly $200 billion each to AI spending this year. Bloomberg data indicates that Nvidia accounts for nearly half of Microsoft's capex and a quarter of Amazon's, creating a massive trickle-down effect for manufacturers like Hon Hai and memory specialists like SK Hynix. Analysts suggest that the broadening demand allows more hardware suppliers to enter the ecosystem, likely leading tech-heavy North Asian markets to continue their outperformance. Surging demand is already manifesting in the financial results of these partners. Samsung's semiconductor division recently reported a 48-fold jump in profit, while SK Hynix saw its quarterly earnings increase fivefold. As the AI buildout evolves from digital generation to real-world deployment, Asia's established infrastructure in advanced hardware and robotics provides a structurally important advantage for implementing the physical AI wave.