China's progress in AI cannot be limited and should not be underestimated, says Nvidia CEO Jensen Huang
Live Events
Amid escalating technological rivalry between the United States and China, Nvidia CEO Jensen Huang underscored China's growing influence in artificial intelligence (AI), describing its progress as undeniable and impressive.In an interview with web portal Stratechery, Huang said the rapid rise of Chinese AI companies such as DeepSeek is impressive."China's doing fantastic; 50% of the world's AI researchers are Chinese and you're not going to hold them back, you're not going to stop them from advancing AI. Let's face it, DeepSeek is deeply excellent work," he said.This was in reference to export controls the US has implemented on advanced chips (such as Nvidia's A100/H100) to prevent uncontrolled AI diffusion to China and other countries.In simple terms, AI diffusion refers to efforts to slow or control the spread of advanced AI technologies (especially foundational models and compute infrastructure) to geopolitical rivals.He said the idea to not have America compete in the Chinese market, where 50% of the developers are, makes no sense from a computing infrastructure and computing architectural perspective. "We ought to go and give American companies the opportunity to compete in China," he said.Huang warned that if US companies don't compete in China, it will in turn allow the Chinese to build a rich ecosystem and new platforms, which would not be American.Earlier this month, Nvidia announced partnerships in the Gulf region , notably with Saudi Arabia and Qatar, to advance AI infrastructure and capabilities. Huang said those countries have an "extraordinary opportunity"."They have an abundance of energy and a shortage of labour, and the potential of their countries is limited by the amount of labour that they have, the amount of people that they have," he said.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Time of India
41 minutes ago
- Time of India
China leaders take reins at TikTok Shop in US as sales miss goal
ByteDance Ltd., TikTok 's parent company, has been replacing US-hired staff near Seattle with leaders connected to China, aiming to replicate its e-commerce success in Asia after sales fell short in America. TikTok Shop initially set a goal to increase its US e-commerce business tenfold last year to $17.5 billion in transaction volume, but the company had to drastically lower that goal, according to people familiar with the plan who spoke on condition of anonymity because they were not authorized to talk publicly. TikTok established its Shop business in the Seattle area near Inc., the online retail giant it was aiming to displace. Meetings that used to be held in English are now often conducted in Mandarin and managers increasingly write in Chinese when communicating on Feishu, ByteDance's internal Slack-like app, with English-speaking staff forced to rely on the built-in translation function. More than 100 TikTok Shop employees in the US have been fired or have left amid confusion between leaders that has worsened the work environment, according to people familiar with the company. The cultural transition taking place in the company coincides with its fight for survival in the US — due mainly to the app's Chinese ties. A national security law passed by Congress last year requires TikTok's US business to be spun off from its Chinese parent company or it will face a ban. Lawmakers warned that TikTok's ties to China pose a threat to the safety and security of American users. President Donald Trump has twice delayed the ban — with legal assurances from his attorney general — and another deadline for divestiture looms later this month, though that might also be extended, Wall Street Journal has reported. ByteDance has said it doesn't intend to sell. The TikTok Shop near Seattle in February began requiring workers to be in the office five days a week for eight hours a day, according to a memo reviewed by Bloomberg. The change is in contrast to some other major tech companies that still offer flexible work schedules, and has been particularly burdensome for employees who often join late-night calls with colleagues in Asia after they leave the office, according to former employees. US-based staff require human resources and manager pre-approval to work from home. The changes were introduced after Bob Kang, China-based global head of TikTok's e-commerce division, visited the office in Bellevue, Washington, earlier this year and found there weren't enough staff pressent on a work day, according to multiple people who spoke on the condition of anonymity for fear of retaliation. Increasing influence Increasing Chinese influence over TikTok's fastest-growing business may raise questions about its previous corporate promise to distance the US operation from China. After Trump initially tried to ban the app during his first term, the company announced a security plan dubbed 'Project Texas' and vowed to wall off the app's US data and operations from any Chinese oversight. TikTok Shop is the biggest source of revenue for the video-sharing app besides advertising, and it has become a major investment area for ByteDance. Adding full-scale commerce to its eye-catching content and popular influencers sets it apart from rivals like Instagram and YouTube. The company still aims to challenge Amazon in major markets. To better compete, TikTok Shop recruited aggressively near Seattle over the past three years, targeting people with experience at Amazon, according to a review of Linkedin profiles and people who worked at both companies. In some corners of TikTok's Bellevue office of roughly 1,000 employees, the workflow felt like a remix of previous Amazon teams, the people said. But since January, growing tension in the teams below Kang and Nico Le Bourgeois, who oversaw TikTok's e-commerce operations in the US, became a distraction for staff who were often unsure about whose orders to follow, the people said. TikTok's uncertain fate in the US also weighed on morale. The company carried out a round of layoffs in April. A second batch followed in May. In the first round, Le Bourgeois was demoted when Mu Qing, a Chinese executive from ByteDance's e-commerce platform Douyin moved to the Seattle area to run TikTok Shop in the US. After the second bout, Mu sent an internal message saying Le Bourgeois was leaving to pursue other opportunities, according to a copy of the message seen by Bloomberg. Those cuts were intended to improve TikTok's 'efficiency,' according to former employees, though it wasn't clear to staff what factors contributed to a worker's efficiency rating. More like Douyin With these changes, ByteDance leaders are bringing in people who are familiar with what worked for the company in China, where Douyin, its TikTok clone for the Chinese market, has evolved into a $490 billion shopping phenomenon. In addition to Mu, who was the head of Douyin's e-commerce, six other leaders with Chinese backgrounds were appointed in April, according to a different internal memo from Kang viewed by Bloomberg. One challenge is that habits of many American users trend toward passive TikTok scrolling as opposed to making purchases in the app. Some US sellers told Bloomberg that they have also been reluctant to invest in the platform, given the possible ban. The final tally for 2024 sales came in at around $9 billion, according to an estimate by Singapore-based consultancy Momentum Works, far below the internal goal of $17.5 billion in transaction volume. A TikTok spokesperson previously called the $17.5 billion internal goal 'inaccurate.' TikTok Shop's US struggles haven't halted the company's global shopping ambitions. ByteDance in 2021 rolled out e-commerce services in countries including Indonesia, Vietnam and the UK. In Southeast Asia, it's already the region's biggest shopping platform after Shopee, according to Momentum Works. Last year, TikTok Shop opened in five countries in Europe, including Germany and Spain. The Europe expansion was delayed because the company first prioritised US growth, Bloomberg reported. A TikTok spokesperson did not respond to an emailed request for comment for this story. This is a crucial month for TikTok in the US. The company will host merchants and creators in Los Angeles next week for a summit featuring some of the new leaders of the e-commerce unit. The current deadline for ByteDance to sell the TikTok's US operation is June 19 and there have been several interested suitors. The company came close to a possible spin-off in April to a consortium of investors that included Oracle Corp., but the deal was scuttled in part because of Trump's trade war with China. Meanwhile, the churn of e-commerce employment continues in the Seattle area. Current and former TikTok Shop employees told Bloomberg that they get hounded by recruiting messages from Temu , another Chinese e-commerce competitor.
&w=3840&q=100)
First Post
42 minutes ago
- First Post
The Deng doctrine: How China weaponises rare earths to gain leverage in trade war with the US
China has signalled for more than 15 years that it was looking to weaponise areas of the global supply chain, a strategy modelled on longstanding American export controls Beijing views as aimed at stalling its rise. read more China has long indicated its intention to weaponise parts of the global supply chain—a strategy now visibly playing out through tighter control of rare earth exports. Modelled on longstanding US export restrictions that Beijing believes are designed to limit its technological rise, China is now turning similar tools to its own advantage. The recent rush by companies to secure export licences for rare earth materials, culminating in a phone call between US President Donald Trump and Chinese President Xi Jinping on Thursday, highlights how Beijing has refined a powerful lever in the ongoing trade war. STORY CONTINUES BELOW THIS AD Industry experts say China may approve more shipments in the near term but it has no plans to dismantle the new system underpinning those approvals. Instead, China's new export licensing regime, closely mirroring the US model grants the government deeper visibility into global supply chokepoints including critical sectors such as electric vehicle motors and precision systems used in missiles. This level of control offers Beijing a potent means to retaliate in the trade dispute while asserting dominance in strategically vital markets. China sharpens rare earth export controls in trade war playbook As relations between the two countries sour and supply chains fracture, both Washington and Beijing appear determined to shift from broad tariffs to more focused, technical barriers—ones that could have lasting implications for industries worldwide. 'China originally took inspiration for these export control methods from the comprehensive U.S. sanctions regime,' Zhu Junwei, a scholar at the Grandview Institution, a Beijing-based think tank focused on international relations told Reuters. 'China has been trying to build its own export control systems since then, to be used as a last resort.' After a phone call with Chinese President Xi Jinping, President Trump said the two leaders were 'straightening out some of the points,' particularly regarding rare earth magnets—key components in electric vehicle (EV) motors and high-tech weaponry. But Trump did not confirm whether Beijing had agreed to speed up export licensing, a sticking point since Washington imposed restrictions on chip design software and jet engines over what it calls China's deliberate slow-walking of approvals. China, which holds a near-monopoly on rare earth magnets, added some of the most advanced types to its export control list in April. The move forces all exporters to seek government licences before shipping these materials, turning a once-obscure division of the commerce ministry—staffed by around 60 people—into a powerful gatekeeper of global manufacturing. STORY CONTINUES BELOW THIS AD The export curbs, part of a broader retaliation package against US tariffs, have had ripple effects well beyond the US. Several European auto parts manufacturers were forced to shut down production lines this week after exhausting their supply of rare earth magnets, underscoring the global reach of Beijing's measures. Though China's commerce ministry has not publicly commented on the issue, analysts say the blanket controls offer Beijing both leverage in its trade war with Washington and a strategic tool to reshape global supply chains in its favour. 'Beijing has a degree of plausible deniability – no one can prove China is doing this on purpose,' Noah Barkin, senior adviser at Rhodium Group, a China-focused U.S. thinktank told Reuters. 'But the rate of approvals is a pretty clear signal that China is sending a message, exerting pressure to prevent trade negotiations with the U.S. leading to additional technology control.' China mines about 70% of the world's rare earths but maintains a near-monopoly on refining and processing, giving it a powerful position in global manufacturing. Even if export approvals accelerate, as U.S. President Donald Trump indicated after a call with President Xi Jinping, Beijing's new licensing system offers it unprecedented visibility into how companies use these critical materials. STORY CONTINUES BELOW THIS AD European and U.S. executives warn that by forcing exporters to apply for licences, China's government can now closely monitor supplier chokepoints in sectors ranging from electric vehicles to advanced weaponry, oversight that other governments lack due to the complexity of global supply chains. Hundreds of Japanese companies are expected to need Chinese export approvals for rare earth magnets in the coming weeks, a person lobbying on their behalf told Reuters. Without timely licences, they risk production disruptions, underscoring how Beijing's new trade tools could reshape access to materials essential to modern industry. 'It's sharpening China's scalpel,' said a US-based executive at a company seeking to piece together an alternative supply chain who sought anonymity. 'It's not a way to oversee the export of magnets, but a way to gain influence and advantage over America.' China's export controls deepen as fears grow over weaponisation of supply chain power Fears that China could weaponise its dominance in critical supply chains first emerged in 2010, when it briefly halted rare earth exports to Japan during a territorial dispute. But those concerns have intensified in recent years as Beijing sharpens its trade tools and broadens export restrictions across strategic sectors. As far back as 1992, former Chinese leader Deng Xiaoping noted, 'The Middle East has oil, China has rare earths.' That sentiment has shaped policy: in 2020, China passed a sweeping Export Control Law allowing it to restrict exports of any items deemed vital to national security, including materials, technology and data. STORY CONTINUES BELOW THIS AD Since then, China has built up its own sanctions arsenal in response to U.S. restrictions, investing heavily in alternative supply chains while tightening its grip on key exports. In 2022, the United States imposed broad curbs on chip and semiconductor tool exports to China, aiming to slow the country's military and AI advancements. But analysts say Beijing has continued to make headway despite those barriers. In retaliation, China has steadily expanded its export controls. Last year it imposed licensing requirements for gallium, germanium, and certain graphite products—vital inputs for defence, electronics, and green technologies. Shipments of these minerals to the U.S. were banned outright in December. Then in February, China added five more metals to its control list. Now, following a phone call between Donald Trump and Xi Jinping, attention has turned to whether China will ease its latest rare earth export curbs. But analysts warn of a lack of transparency. 'It's virtually impossible to know what percentage of requests for non-military end users get approved because the data is not public and companies don't want to publicly confirm either way,' said Cory Combs, an analyst at China-focused consultancy Trivium. STORY CONTINUES BELOW THIS AD The opaqueness of Beijing's process and its expanding powers over chokepoint materials are reinforcing Western concerns that supply chains are becoming geopolitical battlegrounds. With inputs from agencies
Hindustan Times
an hour ago
- Hindustan Times
Thinking AI models collapse in face of complex problems, Apple researchers find
Just days ahead of the much-anticipated Worldwide Developer Conference (WWDC), Apple has released a study titled 'The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity', which saw researchers testing 'reasoning'; AI models such as Anthropic's Claude, OpenAI's o models, DeepSeek R1 and Google's Thinking models to see how far they can scale to replicate human reasoning. Spoiler alert — not as much, as the entire AI marketing pitch, would have you believe. Could this signal what may be in store for Apple's AI conversation ahead of the keynote? The study questions the current standard evaluation of Large Reasoning Models (LRMs) using established mathematical and coding benchmarks, arguing they suffer from data contamination and don't reveal insights into reasoning trace structure and quality. Instead, it proposes a controlled experimental testbed using algorithmic puzzle environments. The limitations of AI benchmarking, and need to evolve, is something we had written about earlier. 'We show that state-of-the-art LRMs (e.g., o3-mini, DeepSeek-R1, Claude-3.7-Sonnet-Thinking) still fail to develop generalizable problem-solving capabilities, with accuracy ultimately collapsing to zero beyond certain complexities across different environments,' the researcher paper points out. These findings are a stark warning to the industry — current LLMs are far from general-purpose reasoners. The emergence of Large Reasoning Models (LRMs), such as OpenAI's o1/o3, DeepSeek-R1, Claude 3.7 Sonnet Thinking, and Gemini Thinking, has been hailed as a significant advancement, potentially marking steps toward more general artificial intelligence. These models characteristically generate responses following detailed 'thinking processes', such as a long Chain-of-Thought sequence, before providing a final answer. While they have shown promising results on various reasoning benchmarks, the capability of benchmarks to judge rapidly evolving models, itself is in doubt. The researchers cite a comparison between non-thinking LLMs and their 'thinking' evolution. 'At low complexity, non-thinking models are more accurate and token-efficient. As complexity increases, reasoning models outperform but require more tokens—until both collapse beyond a critical threshold, with shorter traces,' they say. The illustrative example of the Claude 3.7 Sonnet and Claude 3.7 Sonnet Thinking illustrates how both models retain accuracy till complexity level three, after which the standard LLM sees a significant drop, something the thinking model too suffers from, a couple of levels later. At the same time, the thinking model is using significantly more tokens. This research attempted to challenge prevailing evaluation paradigms, which often rely on established mathematical and coding benchmarks, which are otherwise susceptible to data contamination. Such benchmarks also primarily focus on final answer accuracy, providing limited insight into the reasoning process itself, something that is the key differentiator for a 'thinking' model compared with a simpler large language model. To address these gaps, the study utilises controllable puzzle environments — Tower of Hanoi, Checker Jumping, River Crossing, and Blocks World — and these puzzles allow for precise manipulation of problem complexity while maintaining consistent logical structures and rules that must be explicitly followed. That structure theoretically opens a window, a glance at how these models attempt to 'think'. The findings from this controlled experimental setup reveal significant limitations in current frontier LRMs. One of the most striking observations is the complete accuracy collapse that occurs beyond certain complexity thresholds across all tested reasoning models. This is not a gradual degradation but a sharp drop to near-zero accuracy as problems become sufficiently difficult. 'The state-of-the-art LRMs (e.g., o3-mini, DeepSeek-R1, Claude-3.7-Sonnet-Thinking) still fail to develop generalizable problem-solving capabilities, with accuracy ultimately collapsing to zero beyond certain complexities across different environments,' note the researchers. These results inevitably challenge any notion that the LRMs truly possess generalisation problem-solving skills, required for planning tasks or multi-step processes. The study also identifies a counter-intuitive scaling limit in the models' reasoning effort (this is measured by the inference token usage during the 'thinking' phase), which sees these models initially spend more tokens, but as complexity increases, they actually reduce reasoning effort closer to the inevitable accuracy collapse. Researchers say that 'despite these claims and performance advancements, the fundamental benefits and limitations of LRMs remain insufficiently understood. Critical questions still persist: Are these models capable of generalizable reasoning, or are they leveraging different forms of pattern matching?,' they ask. There are further questions pertaining to performance scaling with increasing problem complexity, comparisons to the non-thinking standard LLM counterparts when provided with the same inference token compute, and around inherent limitations of current reasoning approaches, as well as improvements that might be necessary to advance toward more robust reasoning. Where do we go from here? The researchers make it clear that their test methodology too has limitations. 'While our puzzle environments enable controlled experimentation with fine-grained control over problem complexity, they represent a narrow slice of reasoning tasks and may not capture the diversity of real-world or knowledge intensive reasoning problems,' they say. They do add that the use of 'deterministic puzzle simulators assumes that reasoning can be perfectly validated' at every step, a validation that may not be feasible to such precision in less structured domains. That they say, would restrict validity of analysis to more reasoning. There is little argument that LRMs represent progress, particularly for the relevance of AI. Yet, this study highlights that not all reasoning models are capable of robust, generalisable reasoning, particularly in the face of increasing complexity. These findings, ahead of WWDC 2025, and from Apple's own researchers, may suggest that any AI reasoning announcements will likely be pragmatic. The focus areas could include specific use cases where current AI methodology is reliable (the research paper indicates lower to medium complexity, less reliance on flawless long-sequence execution) and potentially integrating neural models with traditional computing approaches to handle the complexities where LRMs currently fail. The era of Large Reasoning Models is here, but this 'Illusion of thinking' study is that AI with true reasoning, remains a mirage.
