Published on
The world’s most valuable company, the chipmaker Nvidia, priced a $25 billion (€21.5bn) bond offering on Monday, marking its first issuance since 2021 and one of the largest by a technology company this year.
The deal was originally pencilled in at around $20 billion (€17.2bn) but was enlarged after demand ran more than three times the size of the bond, according to a person familiar with the matter cited by Bloomberg.
Investor appetite was the headline of the sale.
Orders reached as high as $85 billion (€73.2bn), allowing Nvidia to upsize the transaction and tighten its borrowing costs in the process.
The timing was also favourable.
The announcement of a US-Iran framework deal to end the conflict in the Middle East steadied credit markets, pushing investment-grade spreads to their narrowest levels since early February, before the Iran war began.
That backdrop helped Nvidia lock in relatively cheap long-term financing.
According to Bloomberg Intelligence analyst Robert Schiffman, inexpensive long-dated debt lowers Nvidia’s weighted average cost of capital and helps bankroll its AI investments without threatening its AA credit rating.
A company spokesperson stated that the proceeds would be used for general corporate purposes, including repaying and refinancing existing notes.
Nvidia last tapped the investment-grade market in June 2021, when it sold $5 billion (€4.3bn) of notes across four maturities, according to a regulatory filing.
The contrast in scale underscores how quickly its financing needs have grown alongside the data centre build-out and increased demand from hyperscalers.
Nvidia joins a queue of technology giants raising vast sums to fund AI infrastructure.
Meta and Oracle have each issued $25 billion (€21.5bn) in bonds this year, while Amazon completed a single $37 billion (€31.8bn) deal, the largest US investment-grade offering of this year before Nvidia’s issuance on Monday.
For Nvidia, the raise also keeps share dilution off the table, giving it greater flexibility as capital commitments mount. The firm has invested $5 billion (€4.3bn) in Intel, pledged up to $10 billion (€8.6bn) to Anthropic and contributed $30 billion (€25.8bn) to OpenAI’s latest funding round.
Nvidia shares closed up 3.5% at $212.45 after the deal, valuing the company at about $5.14 trillion (€4.42tn).
On the other hand, Alphabet, Google’s parent company, opted for equity instead, pricing an upsized $84.75 billion (€73bn) capital raise earlier this month, after originally seeking around $80 billion (€68.9bn), according to a company filing.
The transaction, which includes a $10 billion (€8.6bn) private placement from Berkshire Hathaway, ranks as the largest equity capital raise on record and is intended to fund the group’s AI compute expansion.
Management has guided 2026 capital expenditure to between $180 billion (€155.1bn) and $190 billion (€163.7bn).
However, the equity move came on top of an already heavy borrowing run. According to its own filing, Alphabet raised more than $85 billion (€73.2bn) of debt across six major currencies and markets in the first quarter of 2026, taking its total debt balance above $100 billion (€86.1bn).
That included a US dollar bond round early in the year, leaving Google relying on both debt and equity financing to bankroll its AI ambitions.
Nvidia chief executive officer Jensen Huang, left, and Naver Chairman Lee Hae-jin appear at Naver’s Vision Studio at its 1784 headquarters in Seongnam, south of Seoul, on Monday. Photo courtesy of Naver
June 12 (Asia Today) — OpenAI chief executive officer Sam Altman is expected to visit Naver on Monday, a week after Nvidia chief executive officer Jensen Huang met with the South Korean technology company’s leadership.
Industry sources said Altman plans to visit South Korea during a two-day trip beginning Sunday and meet officials from major Korean companies, including Naver, Kakao and Samsung Electronics, to discuss artificial intelligence cooperation.
The expected Naver visit comes shortly after Huang visited Naver’s 1784 headquarters in Seongnam, south of Seoul, on Monday and met with Lee Hae-jin, Naver’s founder and chairman.
The back-to-back visits by two of the most influential figures in the global AI industry are drawing renewed attention to Naver’s role in the sector. Industry officials said OpenAI may be interested in Naver’s data, service ecosystem and experience operating consumer platforms at scale.
Naver has built large user data assets through search, shopping, content, community, mapping and reservation services. As competition in generative AI increasingly depends on access to high-quality data, those assets are viewed as one of Naver’s main strengths.
Naver recently outlined what it calls a “product-native LLM” strategy, saying it will optimize AI for specific services such as search, shopping, maps and reservations rather than rely only on a single general-purpose model.
The company develops its own AI models and also operates a creator ecosystem of about 20 million people, with more than 630 million pieces of content produced annually. It also owns large-scale data center infrastructure and runs the services where AI can be applied directly to users.
A Naver official said competition in AI is changing quickly.
“In the past, the key was developing a better model,” the official said. “Now, the ability to secure high-quality data, service experience and the infrastructure to support them is emerging as a decisive factor.”
Naver is also seeking to expand AI search into agentic AI services, in which AI does not simply answer a user’s question but can help complete tasks such as reservations and purchases.
Global interest in Naver also grew after Huang’s visit this week. During his meeting with Lee, the Nvidia chief described Naver as a “world-class AI company.”
Huang cited possible cooperation with Naver in several areas, including participation in Nvidia’s Nemotron Alliance, the development of AI factories and robotics. He said Naver was selected because it has world-class cloud technology and AI talent.
Industry officials said Nvidia appears to view Naver as an AI infrastructure partner, while OpenAI may see potential in Naver’s data and service ecosystem.
“Naver’s combined strength in AI models, data, services and infrastructure is attracting attention from global AI companies,” one industry official said.
— Reported by Asia Today; translated by UPI
© Asia Today. Unauthorized reproduction or redistribution prohibited.
Original Korean report: https://www.asiatoday.co.kr/kn/view.php?key=20260612010004347
SpaceX is set for the largest stock market debut ever.
Elon Musk’s rocket company begins trading on the Nasdaq on Friday under the ticker SPCX. The company priced its shares at $135 each, raising $75 billion (€64.5bn) and valuing the business at $1.75 trillion (€1.5trn) in the biggest stock market flotation on record.
The deal would comfortably eclipse Saudi Aramco’s previous record of $29.4bn, set in 2019 and later increased through an overallotment option.
SpaceX made an unusually strong push to attract retail investors, including those in Europe. According to Bloomberg, individual investors placed roughly $100bn (€86.6bn) in orders through trading platforms including Robinhood, Fidelity and SoFi during the IPO process.
That demand alone exceeded the company’s $75bn (€64.5bn) fundraising target, underscoring the level of interest from smaller investors ahead of the stock market debut.
Yet beneath the hype, several warning lights are flashing. Here are five risks investors should weigh before the SpaceX IPO goes live.
At a valuation of $1.75tn (€1.5trn), investors would be valuing SpaceX at roughly 94 times its annual revenue, which was $18.7bn (€16.1bn) in 2025. By comparison, Nvidia — one of the market’s most highly valued technology companies — trades at less than a quarter of that level.
The investment research firm Morningstar, which values the company at $780bn (€675bn), called it “significantly overvalued” while Goldman Sachs data suggests sustaining the share price would require revenues above $100bn (€86.6bn) by 2030, implying a compound annual growth of more than 40%.
History offers a note of caution. Research by University of Florida professor Jay Ritter, often referred to as “Mr IPO”, found that while IPOs between 2012 and 2021 rose an average of 23.6% on their first day of trading, they returned just 10.6% over the following three years.
SpaceX’s expected inclusion in major stock indexes has become a point of controversy. Investment officials from four large US states have urged Nasdaq and FTSE Russell to explain recent rule changes that could accelerate the company’s entry into widely tracked benchmarks.
Critics argue the move could expose passive investors to a highly valued stock sooner than expected, while the index providers say the changes reflect broader market developments.
The debate matters because relatively few SpaceX shares will initially be available for trading. Although SpaceX is valued at $1.75tr (€1.5trn), only around 3% to 4% of its shares will initially be available for public trading.
That means the company’s market value will be determined by trading in a relatively small portion of its equity. Reports suggest more than 75% of the $75bn (€64.5bn) offering has already been allocated to existing investors and insiders, leaving fewer shares available on the open market.
According to Morningstar, the limited float and strong demand for artificial intelligence-related stocks could help support the share price in the early stages of trading, even if the company is valued above what the research firm considers fair value. The firm argues that a clearer picture of investor demand may emerge once lock-up restrictions expire and more shares become available for trading.
Some analysts, however, believe the limited float could continue to support the stock. Estimates suggest between $22 billion (€19bn) and $27 billion (€23.4bn) of passive investment could flow into SpaceX once it joins the Nasdaq 100, creating additional demand from index-tracking funds.
SpaceX’s financial results may also give investors pause.
The prospectus shows that the company is growing rapidly but still losing money.
The company owns the Starlink satellite internet service, which generates most of its revenue and is its only profitable business. It also owns the artificial intelligence company xAI, which merged with SpaceX in February.
According to the filing, SpaceX carried an accumulated deficit of $41.3bn (€35.76bn) as of 31 March and reported a net loss of $4.27bn (€3.7bn) in the first quarter of 2026.
This compares with $528mn (€457mn) in the same period a year earlier.
Much of the recent loss stems from xAI. According to SpaceX’s IPO filing, the AI business recorded an operating loss of about $6.4 billion (€5.5bn) in 2025. The filing also showed xAI spent heavily in the opening months of 2026 as it expanded its AI infrastructure.
Morningstar argues the AI unit “poses a material threat of value destruction”, noting that Grok has yet to win meaningful market share against rival chatbots.
Supporters counter that the losses are a choice, not a structural flaw.
Revenue climbed 33% to $18.7bn (€16.2bn) in 2025, up from $14.1 billion (€12.2bn) a year earlier. The underlying launch and satellite business was profitable as recently as 2024. The deficits largely reflect heavy investment in AI infrastructure, spending that supporters say is already beginning to be offset by new compute contracts.
Supporters argue investors are paying for future growth rather than current profits.
Starlink remains the company’s main source of revenue, while its artificial intelligence business is expected to play a larger role in the years ahead.
Bulls also point to SpaceX’s dominant position in rocket launches and satellite communications, arguing the company is uniquely placed to benefit from growing demand for connectivity, computing power and AI infrastructure.
SpaceX conducts more rocket launches annually than the rest of the world combined and counts over nine million Starlink subscribers, but its newest growth driver is the AI data-centre business acquired through the xAI merger.
Last Friday, Google agreed to pay SpaceX $920 million (€796.6mn) per month for compute capacity at xAI data centres, in a 32-month deal running from October 2026 through June 2029, and covering access to roughly 110,000 Nvidia GPUs.
That followed a May agreement under which Anthropic pays $1.25 billion (€1.08bn) a month to rent the entire output of the Colossus 1 data centre until May 2029, putting combined annualised compute revenue at around $26 billion (€22.5bn).
Bulls argue this contracted income, won in under four months, shows how quickly the company can monetise its infrastructure. Sceptics note that both contracts carry 90-day termination clauses after December 2026, and that Google itself has framed the arrangement as “bridge capacity” rather than a permanent commitment.
SpaceX’s success is closely tied to Elon Musk, whose profile and track record have helped attract investors, customers and business partners. That creates what investors call “key-person risk” — concerns about how the company would fare if he were no longer leading it.
The company’s governance structure reinforces that dependence. Musk’s super-voting Class B shares give him around 85% of voting power, leaving outside shareholders with little influence over major corporate decisions. In practice, that means no one but Musk himself can determine whether he remains chief executive.
Critics also point to SpaceX’s incorporation in Texas, where only investors holding at least 3% of shares can bring derivative lawsuits. The Danish academic pension fund AkademikerPension has blacklisted the stock, describing the governance structure as “catastrophic”.
Supporters argue that dual-class share structures are common among US technology firms, including Meta and Alphabet. They say concentrated voting control allows founders to pursue long-term goals without pressure from short-term investors.
Musk’s prominence also brings political risk. US Senator Elizabeth Warren has urged the Securities and Exchange Commission to scrutinise the listing, warning that future index inclusion could expose millions of passive investors to the stock without them actively choosing it.
Others note that the SEC completed its review faster than expected, allowing the IPO process to move ahead without delay and suggesting regulators see no immediate obstacle to the listing.
Disclaimer: This information does not constitute financial advice, always do your own research on top to ensure it’s right for your specific circumstances. Also remember, we are a journalistic website and aim to provide the best guides, tips and advice from experts. If you rely on the information here, then you do so entirely at your own risk.
Nvidia Chief Executive Jensen Huang, left, and Naver founder and board Chairman Lee Hae-jin greet attendees at Naver’s 1784 headquarters in Seongnam, south of Seoul, on Monday. Photo by Asia Today
June 8 (Asia Today) — Naver said Monday it will work with Nvidia to build a gigawatt-scale artificial intelligence factory, starting from its hyperscale data center in Sejong.
Naver founder and board Chairman Lee Hae-jin and Chief Executive Choi Soo-yeon met Nvidia Chief Executive Jensen Huang at Naver’s 1784 headquarters in Seongnam, south of Seoul, to discuss a joint business road map and global expansion strategy.
Huang greeted employees and visitors at the building, saying, “I love Naver.” He also joined a Naver Webtoon event and wrote, “Don’t worry! I have GPUs!” in a blank space on a display.
Naver and Nvidia said they agreed to pursue a joint project to build a large-scale global AI factory. The partnership goes beyond technology cooperation, covering demand development, investment and infrastructure construction across the value chain.
Naver will participate as a core partner sharing business results and risks.
The project will be based at Gak Sejong, Naver’s hyperscale data center. Naver plans to begin operating 55 megawatts of infrastructure in the first half of 2027, expand to 100 megawatts later that year and reach 200 megawatts in 2028. The company ultimately aims to build gigawatt-scale AI infrastructure.
Naver plans to use Gak Sejong to serve AI demand not only in South Korea but also in Asia, the Middle East and Europe.
The cooperation centers on combining Naver’s data-center and GPU cluster operation capabilities with Nvidia’s DSX platform. Nvidia DSX integrates chips, servers, software and data-center operating technologies for AI factories. The platform is designed to lower AI model training and inference costs and speed up infrastructure deployment.
Naver plans to use the technology to expand AI infrastructure services for companies, governments and industrial clients.
The companies also plan to broaden technical cooperation. Naver has been improving its HyperCLOVA X AI model by using Nvidia’s open large language model Nemotron. It is also working to develop a “Seoul world model” by combining Nvidia’s Cosmos world foundation model with Naver’s street-view and spatial modeling technologies.
The cooperation is expected to expand into physical AI, robotics and digital twins.
The meeting marked another step in the companies’ existing partnership. Lee and Huang met last year during the Asia-Pacific Economic Cooperation summit in Gyeongju to discuss cooperation on physical AI platforms.
Naver 1784 is considered a showcase for robotics, digital twins and cloud technologies. Huang was expected to review areas for expanded cooperation during his visit.
Naver is seeking to move beyond its role as an internet services company and become a global AI infrastructure provider. Huang recently introduced Naver Cloud as a key partner in the global AI ecosystem during Nvidia GTC Taipei 2026. Naver said it plans to accelerate its sovereign AI and AI data-center businesses through cooperation with Nvidia.
Naver shares also rose Monday. The stock closed at 279,000 won, or about $181, up 9.20% from the previous trading session, according to the Korea Exchange. Market analysts attributed the gain to investor expectations for the large-scale AI factory project and Naver’s global AI infrastructure expansion.
— Reported by Asia Today; translated by UPI
© Asia Today. Unauthorized reproduction or redistribution prohibited.
Original Korean report: https://www.asiatoday.co.kr/kn/view.php?key=20260608010002635
NVIDIA CEO Jensen Huang delivers his keynote speech as part of the COMPUTEX 2026 AI exhibition in Taipei, Taiwan, 01 June 2026. Jensan Huang officially announced NVIDIA’s new products and various AI integration. Photo by RITCHIE B. TONGO / EPA
June 2 (Asia Today) — Naver Cloud is moving to expand its presence in the global artificial intelligence infrastructure market through a deeper partnership with Nvidia.
The company aims to combine its HyperCLOVA X large-scale AI model and sovereign AI capabilities with Nvidia’s AI infrastructure platform to become a key player in the era of AI factories.
Industry officials said Tuesday that Naver Cloud CEO Kim Yu-won attended the Nvidia Cloud Partner Summit in Taiwan and outlined the company’s strategic cooperation with Nvidia.
The partnership is drawing attention because it goes beyond a simple graphics processing unit supply arrangement and extends across infrastructure, AI models and services.
Nvidia CEO Jensen Huang introduced Naver Cloud as a major AI-native cloud partner in the global AI ecosystem during his keynote speech at GTC Taipei 2026 on Monday.
The two companies also plan to expand cooperation in large language models. Naver Cloud plans to use Nvidia’s open large language model technology, Nemotron 3 Ultra, to advance HyperCLOVA X. The companies also plan to jointly study model optimization and core technologies.
Cooperation will also continue in physical AI. In March, Naver Cloud unveiled the Seoul World Model, a digital recreation of Seoul built with Nvidia’s Cosmos physical AI platform.
The Seoul World Model was trained on South Korean map data and 1.2 million panoramic images collected across Seoul, allowing it to reproduce real road environments and spatial structures.
Naver Cloud plans to target the global AI market by emphasizing its full-stack capabilities, which combine its own AI models and cloud infrastructure. The company also plans to expand sovereign AI projects that protect national data sovereignty by working with governments and local companies.
Naver board chair Lee Hae-jin and Huang are expected to meet soon in South Korea and disclose specific plans for the AI factory project.
“The AI industry paradigm is shifting from models to inference-focused AI factories that can operate large-scale infrastructure reliably,” Kim said. “Our cooperation with Nvidia is not a simple supply relationship but a strategic decision to expand the global AI ecosystem together.”
— Reported by Asia Today; translated by UPI
© Asia Today. Unauthorized reproduction or redistribution prohibited.
Original Korean report: https://www.asiatoday.co.kr/kn/view.php?key=20260602010000541
Nvidia CEO Jensen Huang, right, visits the SK hynix booth at Computex 2026 with SK Group Chairman Chey Tae-won on Tuesday. Photo courtesy of SK hynix
June 2 (Asia Today) — Nvidia CEO Jensen Huang visited the SK hynix booth at Computex 2026 in Taipei on Tuesday, meeting SK Group Chairman Chey Tae-won for a second straight day as the companies deepen their artificial intelligence partnership.
Huang, who met privately with Chey on Monday, examined SK hynix’s major memory products and wrote “Please Make More” on an HBM4E wafer displayed at the booth.
Chey also signaled that SK plans to expand production. He said the group aims to double wafer production capacity within five years as demand for memory chips is expected to surge.
Huang toured the booth with Chey and SK hynix executives. He signed the HBM4E wafer with the message “Please Make More” and wrote “LOVE SOCAMM” on a 192GB SOCAMM product.
SK hynix currently supplies Nvidia with its latest high-bandwidth memory, including sixth-generation HBM4, as well as high-performance low-power LPDDR5X memory. Huang said in his GTC Taipei keynote Monday that Nvidia will begin full-scale production of its next-generation AI accelerator, Vera Rubin, in the second half of this year.
As AI demand increases and memory supply shortages deepen, Chey said SK is moving quickly to expand production.
“The memory bottleneck is expected to continue until 2030,” Chey told reporters at the SK hynix booth. “We are pushing forward at full speed to expand production capacity.”
“Building new memory fabs requires enormous investment and takes at least three years,” he said. “Despite these challenges, we plan to double wafer production capacity over the next five years.”
It was the first time SK Group publicly presented a specific goal of doubling its overall production capacity within five years. SK hynix is making large-scale investments to strengthen production capacity, including projects at its M15X and P&T7 facilities in Cheongju, the Yongin semiconductor cluster and an advanced packaging plant in the United States.
— Reported by Asia Today; translated by UPI
© Asia Today. Unauthorized reproduction or redistribution prohibited.
Original Korean report: https://www.asiatoday.co.kr/kn/view.php?key=20260602010000823
Published On 1 Jun 20261 Jun 2026
Nvidia is set to bring artificial intelligence to laptop and desktop computers with brands like Microsoft and Dell later this year as the US tech giant broadens its AI presence.
The Santa Clara, California-based AI chipmaker unveiled on Monday at its annual Nvidia GTC event in Taipei new powerful chips that would bring advanced AI functions to laptops and desktop computers.
list of 4 itemsend of list
CEO Jensen Huang said that the new development is “going to reinvent the PC [personal computer]”.
The changes come amid three years of collaboration between Microsoft and Nvidia and pit the latter against companies like chipmaker Advanced Micro Devices and personal computer brands Intel and Apple.
“This is going to be the new PC,” Huang said as he unveiled Nvidia’s RTX Spark superchip — which combines CPU, or central processing unit, and GPU, or graphics processing unit, capabilities — that would power new Windows laptop and desktop computer models in what the company called “AI personal computers”, expected to debut in the fall of this year.
The chip, developed with Taiwan’s MediaTek, will be in compact desktops from Dell, HP, Lenovo, ASUS, Microsoft Surface and MSI, with models from Acer and GIGABYTE to follow.
Nvidia, which is already the world’s most valuable company, said the reinvention will be for creating and gaming.
“When it has an autonomous [AI] agent, an agent that’s helping you, that understands you, you could talk to it. It could look at you. You could ask it to read files, go help you do some research. It could do a lot more,” Huang said.
Microsoft said in a separate statement that the personal computers running on Nvidia’s RTX Spark superchips would be able to support “highly capable AI models” and complex workloads. With the new superchips, these personal computers can run AI agents locally, Nvidia said.
“This is the first across the lineup of PC reinvention for 40 years,” Huang said.
Nvidia’s move is significant at a time when demand is growing for the use of personal AI agents, said Lian Jye Su, chief analyst at the technology research and advisory group Omdia.
“For consumers, it means more choices, which is always a good thing,” Su said.
Neil Shah, analyst and co-founder of Counterpoint Research, described Nvidia’s announcement as a move that’s “revolutionising how PCs would look like in the next 10 years”.
The new laptops and desktop computers “will drive agentic AI applications in every home”, Shah said, with an aim of having an “AI supercomputer” in each household.
Also during Monday’s speech, Nvidia’s Huang said its new Vera CPUs for data centres are in full production and are “going to be our new major growth driver” on the boom of AI agents, with early customers including Anthropic, OpenAI and SpaceXAI.
Huang also revealed a humanoid robot reference design that could act as a blueprint for future research, especially within the higher education sector. Nvidia said its “Isaac GR00T” stands nearly 1.83 metres (6 feet) tall and has the humanoid chassis of Chinese robot maker Unitree’s H2. It is equipped with five-fingered dexterous hands, made by Singapore-based robotics startup Sharpa, that are capable of finely controlled movements.
Reception for AI PCs has been mixed so far. HP reported last week that the devices helped prop up quarterly sales, but Dell said earlier this year that demand had fallen short of initial expectations. Qualcomm, looking to capitalise on AI demand, has also been offering AI PCs with Microsoft.
On Wall Street, Nvidia stock rose nearly 4 percent on the news in midday trading. Microsoft ticked up 2.5 percent and Dell surged 9.3 percent. Competitors AMD and Intel, on the other hand, are on the decline. AMD is down 0.1 percent from the market open, and Intel is down by 2.5 percent.
BING-JHEN HONG/iStock Editorial via Getty Images
Nvidia’s (NVDA) CEO Jensen Huang said the company plans to invest about $150B a year in Taiwan while announcing a new expanded campus in Taipei.
“Four years ago, five years ago, Nvidia was spending about 10, 15 billion dollars a year in Taiwan. Now
Earnings Call Insights: NVIDIA (NVDA) Q1 fiscal 2027
“This was an extraordinary quarter, demand has gone parabolic. The reason is simple, agentic AI has arrived. AI can now do productive and valuable work. Tokens are now profitable, so model makers are in a race
Seeking Alpha’s Disclaimer: This article was automatically generated by an AI tool based on content available on the Seeking Alpha website, and has not been curated or reviewed by humans. Due to inherent limitations in using AI-based tools, the accuracy, completeness, or timeliness of such articles cannot be guaranteed. This article is intended for informational purposes only. Seeking Alpha does not take account of your objectives or your financial situation and does not offer any personalized investment advice. Seeking Alpha is not a licensed securities dealer, broker or US investment adviser or investment bank.
The latest developments surrounding Nvidia’s H200 chip sales to China highlight the growing complexity of the technological rivalry between the United States and China. Although Washington has reportedly approved several major Chinese firms to purchase Nvidia’s advanced artificial intelligence chips, no deliveries have taken place so far.
The situation reflects how geopolitical competition is increasingly disrupting even officially approved commercial agreements in the semiconductor sector.
Nvidia, the world’s leading artificial intelligence chip manufacturer, now finds itself caught between United States export control policies and China’s push for technological self reliance.
The H200 is Nvidia’s second most powerful artificial intelligence chip and is designed for advanced AI model training and data center operations.
The chip is particularly valuable for companies developing large language models, cloud computing systems, and next generation AI applications.
Before export restrictions tightened, Nvidia dominated China’s advanced AI chip market with an estimated market share of around 95 percent.
China also represented a major source of revenue for Nvidia, making access to the Chinese market strategically important for the company’s long term growth.
According to reports, the United States Commerce Department approved around ten Chinese firms to purchase H200 chips.
These reportedly include major Chinese technology companies such as:
Several distributors were also reportedly approved, including:
Under the licensing terms, each approved customer could reportedly purchase up to 75,000 chips.
However, despite these approvals, no actual sales or deliveries have yet been completed.
The delays appear to stem from concerns on both the United States and Chinese sides.
Chinese authorities reportedly fear that reliance on Nvidia chips could undermine Beijing’s efforts to strengthen its domestic semiconductor industry.
China has invested heavily in local AI chip development, particularly through companies such as Huawei.
Beijing increasingly sees semiconductor self sufficiency as a national security priority amid escalating technological competition with Washington.
There are also concerns within China regarding supply chain security and possible vulnerabilities linked to imported American technology.
Recent Chinese regulations aimed at reducing foreign dependence in critical technology sectors have reportedly intensified scrutiny of these chip purchases.
The United States has simultaneously imposed strict export control requirements on advanced semiconductor sales to China.
Chinese buyers must reportedly prove that the chips will not be used for military purposes and that adequate security procedures are in place.
Nvidia must also satisfy inventory and compliance conditions under American export laws.
Additionally, reports suggest the Trump administration negotiated an unusual arrangement in which the United States would receive a portion of revenue generated from the chip sales. This reportedly requires the chips to pass through American territory before shipment to China.
Such conditions have further complicated the transaction process.
Nvidia Chief Executive Officer Jensen Huang has emerged as a key figure in efforts to preserve Nvidia’s access to the Chinese market.
Huang reportedly joined President Donald Trump during a diplomatic visit linked to talks with Chinese President Xi Jinping.
His participation underscores the economic significance of the semiconductor dispute and the importance of China to Nvidia’s business strategy.
Huang has repeatedly warned that export controls risk permanently weakening Nvidia’s position in China while encouraging Chinese firms to accelerate domestic alternatives.
The Nvidia dispute reflects a broader struggle between the United States and China over technological dominance in artificial intelligence.
Washington increasingly views advanced semiconductor technology as a strategic national security asset. American policymakers fear that unrestricted access to advanced AI chips could strengthen China’s military and technological capabilities.
China, meanwhile, sees semiconductor independence as essential to reducing vulnerability to foreign pressure and sanctions.
As a result, both sides are attempting to balance economic interests with long term strategic competition.
The uncertainty surrounding Nvidia’s China business could have major implications for the global artificial intelligence industry.
If Chinese companies lose access to Nvidia chips, they may accelerate investment in domestic alternatives, potentially reshaping the global semiconductor market over time.
At the same time, restrictions on AI chip trade risk fragmenting the global technology ecosystem into competing American and Chinese spheres.
This could reduce international collaboration, disrupt supply chains, and intensify geopolitical competition over emerging technologies.
Despite current delays, neither the United States nor China appears willing to completely sever technological and commercial ties.
However, the Nvidia case demonstrates that semiconductor trade between the two powers is becoming increasingly politicized and strategically sensitive.
The future of AI competition may ultimately depend not only on innovation, but also on which country can build the most resilient and independent technology ecosystem.
For Nvidia, maintaining its position between the world’s two largest economies will likely remain one of its greatest strategic challenges.
The stalled Nvidia H200 deal illustrates how deeply geopolitical tensions now shape the global technology industry.
Although the United States has approved limited chip exports to China, political distrust, national security concerns, and strategic competition continue to obstruct implementation.
As artificial intelligence becomes central to economic and military power, semiconductor trade is no longer simply a commercial issue. It has become a defining arena in the broader contest between Washington and Beijing for technological leadership in the twenty first century.
With information from Reuters,
Sundry Photography
The U.S. has cleared around 10 Chinese firms to buy Nvidia’s (NVDA) second-most powerful AI chip, the H200, but not a single delivery has been made so far, leaving a major technology deal in limbo as CEO Jensen Huang seeks a breakthrough in China
LG AI Research head Lim Woo-hyung (L) speaks with NVIDIA Vice President Bryan Catanzaro at the company’s head office in Seoul on Tuesday. Photo by LG Group
SEOUL, April 22 (UPI) — South Korea’s LG AI Research said that it has agreed to strengthen cooperation with NVIDIA to develop next-generation AI technologies and expand the ecosystem of its flagship AI model, EXAONE.
Toward that goal, LG AI Research’s chief Lim Woo-hyung met with NVIDIA Vice President Bryan Catanzaro, who visited Korea to attend the NVIDIA Nemotron Developer Days Seoul 2026.
The two companies have collaborated before. LG AI Research said that it has leveraged datasets of NVIDIA’s Nemotron open ecosystems to develop and upgrade its EXAONE models.
“Purpose-built, domain-specific models unlock the full value of AI by using culture- and language-specific data aligned with what makes nations and industries unique,” Catanzaro said in a statement.
“By integrating the LG AI Research EXAONE platform with NVIDIA Nemotron, organizations can create high-quality local models that advance sovereign AI initiatives-opening the door to new business opportunities and enhanced social services.”
Lim stressed that NVIDIA has been a key partner throughout the development of EXAONE.
“We will expand our collaboration with NVIDIA beyond research into a broader innovation ecosystem to deliver tangible sovereign AI outcomes that can be realized across industries,” he said.
As one of the leaders in South Korea’s sovereign AI project, LG Group has recently sought to accelerate the conglomerate’s AI transformation.
Earlier this month, for example, Chairman Koo Kwang-mo flew to Silicon Valley to meet with chiefs of global tech companies Palantir Technologies and Skild AI.
The share price of LG Corp., the holding company of LG Group, gained 0.95% on the Seoul bourse Wednesday.
