AI chatbots using reason emit more carbon than those responding concisely, study finds
'The environmental impact of questioning trained (large-language models) is strongly determined by their reasoning approach, with explicit reasoning processes significantly driving up energy consumption and carbon emissions,' first author Maximilian Dauner, a researcher at Hochschule München University of Applied Sciences, Germany, said.
'We found that reasoning-enabled models produced up to 50 times more (carbon dioxide) emissions than concise response models,' Dauner added.
The study, published in the journal Frontiers in Communication, evaluated how 14 large-language models (which power chatbots), including DeepSeek and Cogito, process information before responding to 1,000 benchmark questions — 500 multiple-choice and 500 subjective.
Each model responded to 100 questions on each of the five subjects chosen for the analysis — philosophy, high school world history, international law, abstract algebra, and high school mathematics.
'Zero-token reasoning traces appear when no intermediate text is needed (e.g. Cogito 70B reasoning on certain history items), whereas the maximum reasoning burden (6.716 tokens) is observed for the Deepseek R1 7B model on an abstract algebra prompt,' the authors wrote.
Tokens are virtual objects created by conversational AI when processing a user's prompt in natural language. More tokens lead to increased carbon dioxide emissions.
Chatbots equipped with an ability to reason, or 'reasoning models', produced 543.5 'thinking' tokens per question, whereas concise models — producing one-word answers — required just 37.7 tokens per question, the researchers found.
Thinking tokens are additional ones that reasoning models generate before producing an answer, they explained.
However, more thinking tokens do not necessarily guarantee correct responses, as the team said, elaborate detail is not always essential for correctness.
Dauner said, 'None of the models that kept emissions below 500 grams of CO2 equivalent achieved higher than 80 per cent accuracy on answering the 1,000 questions correctly.' 'Currently, we see a clear accuracy-sustainability trade-off inherent in (large-language model) technologies,' the author added.
The most accurate performance was seen in the reasoning model Cogito, with a nearly 85 per cent accuracy in responses, whilst producing three times more carbon dioxide emissions than similar-sized models generating concise answers.
'In conclusion, while larger and reasoning-enhanced models significantly outperform smaller counterparts in terms of accuracy, this improvement comes with steep increases in emissions and computational demand,' the authors wrote.
'Optimising reasoning efficiency and response brevity, particularly for challenging subjects like abstract algebra, is crucial for advancing more sustainable and environmentally conscious artificial intelligence technologies,' they wrote. PTI KRS KRS MPL
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Mint
17 hours ago
- Mint
AI Boom Seen Driving Next Decade of Emerging Markets Performance
(Bloomberg) -- Emerging-market funds are pivoting to capture the artificial intelligence craze, with some investors predicting that booming technology spending will drive returns for years to come. Encouraged by the success of Chinese AI developer DeepSeek and Asia's powerhouse semiconductor firms, asset managers like AllSpring Global Investments and GIB Asset Management are concentrating more of their portfolio in AI stocks. That's been a winning trade, with AI companies being the six biggest contributors to the rally in Bloomberg's EM stocks index this year. 'This trend could last for the next 10 to 20 years,' said Alison Shimada, head of total emerging markets equity at AllSpring, which oversees $611 billion. 'The impact on local populations within EM will be transformational.' While much of the AI investment frenzy has focused on a handful of Silicon Valley firms, EM companies that can harness the technology or supply crucial components are benefitting. AI servers, for example, have become the main growth driver for Taiwan's Hon Hai Precision Industry Co., which is known as Foxconn. The top contributors to Bloomberg's EM stock index this year are Taiwan Semiconductor Manufacturing Co., Tencent Holdings Ltd., Alibaba Group Holding Ltd., Samsung Electronics Co., SK Hynix Inc. and Xiaomi Corporation, together accounting for 37% of the index's rally. Emerging-market stocks that are highly exposed to AI have even outperformed the so-called Magnificent Seven megacap tech firms so far this year, according to equities strategists at Citigroup Inc. 'You cannot invest in emerging markets without having a sanguine and optimistic view of what this AI story can evolve into from a corporate earnings perspective,' said Kunal Desai, London-based co-portfolio manager for global emerging markets equities at GIB Asset Management. Desai said that Taiwan and South Korea will be 'central drivers' of the EM market story over the next two to three years, with Malaysia, China, India, parts of Latin America and the Middle East seeing 'disproportionate gains' due to their exposure to AI data and applications. His fund has invested in AI stocks during recent market dips, predicting that a third of emerging market returns will come from AI-related stocks in the coming years. There are signs that the momentum will continue as AI adoption accelerates across segments including cloud computing and electrical vehicles. The average estimate of forward 12-month earnings for EM tech stocks has increased 15% since the start of the year, compared to 6% for EM stocks overall. 'The share of AI contribution from the performance standpoint will only grow from here,' said Xingchen Yu, an emerging markets strategist at UBS Global Wealth Management. 'The rise of AI and tech is creating a new layer of secular growth, especially in North Asia.' The AI revolution could help EM stocks overcome a key obstacle: earnings performance. Company results have lagged forecasts every quarter since early 2022, with MSCI EM Index companies collectively missing profit expectations by more than 12%, according to data compiled by Bloomberg. But firms in the AI-heavy information-technology sector have consistently met earnings projections since the fourth quarter of last year, boosting investor confidence. 'This sector has been expected to grow explosively and will continue to do so in the future,' said Young Jae Lee, senior investment manager at Pictet Asset Management Ltd. 'AI will continue to be a key sector within emerging markets.' --With assistance from Jorgelina do Rosario. More stories like this are available on
Mint
19 hours ago
- Mint
Politics beats the market in Trump's pay-for-play chip scheme
The Trump administration's recent reversal of export restrictions on certain Nvidia and AMD's AI chips to China marks a major shift in U.S. policy and has raised legal concerns. But the deeper risks are political and economic—and warrant scrutiny. His administration decided in July to re-allow Nvidia and AMD to sell their H20 and MI308 chips to China—conditional on a 15% revenue remittance to the U.S. government. Export controls on AI chips were initially implemented by the Biden administration in 2022 to curb China's access to advanced processors critical for artificial intelligence and supercomputing, citing national security concerns. But Nvidia's H20 chip was specifically designed for the Chinese market. Even though it is reportedly incapable of training large AI models, the H20 chip is effective for inference tasks—allowing AI systems to respond to queries based on pre-trained data. Chinese engineers have leveraged the H20 chips for applications using open-source models such as DeepSeek and Alibaba's Qwen, which are increasingly popular in China's AI ecosystem. Trump continued these controls—until recently. Critics, such as Rep. Don Bacon (R., Neb.) and Liza Tobin, who served as China director at the National Security Council under the first Trump and Biden administrations, have argued that the sale of these chips, even under revenue-sharing constraints, undermines the strategic intent of the original export controls. Rep. Raja Krishnamoorthi (D., Ill.) likened it to gambling with national security. The Trump administration's reversal appears motivated by revenue generation. Bernstein Research estimates that Nvidia could sell approximately 1.5 million H20 chips in China in 2025, generating $23 billion in revenue; a 15% cut would yield more than $3 billion for the U.S. government. It aligns with Trump's broader strategy of using tariffs and trade deals to bolster government income. However, this revenue-centric approach raises concerns about the erosion of principled policymaking. The arrangement resembles a 'pay-to-play" scheme, where export licenses are granted in exchange for financial contributions. Legal scholars argue that this violates the Export Clause of the U.S. Constitution, which prohibits taxes or duties on exports. Moreover, the Export Controls Reform Act of 2018—signed by Trump himself—explicitly forbids charging exporters for licenses. The deal with Nvidia and AMD may also reflect a broader diplomatic calculus. China dominates the global supply chain for rare earth materials, which are essential for military technologies like guided missiles, fighter jets, and radar systems. These materials are also used in manufacturing key components for smartphones and electric vehicles—including batteries, touch screens, and camera lenses. The U.S. is investing in domestic mining and processing of rare earth minerals, such as at the Mountain Pass mine in California, which is the only rare-earth mining and processing facility in the U.S. But it remains heavily reliant on Chinese exports. By easing chip restrictions, the Trump administration may be signaling goodwill in hopes of securing a stable supply of rare earths. This strategic compromise, however, risks emboldening China. If Beijing perceives U.S. export controls as negotiable or monetizable, it may be less inclined to make concessions in other areas of trade or security. The precedent set by this deal could weaken the credibility of future U.S. restrictions, making it harder to enforce technology bans or secure allied cooperation. And then there are the corporate interests. The role of Nvidia CEO Jensen Huang in shaping this policy shift cannot be overlooked. Huang reportedly argued that restricting Nvidia's access to the Chinese market would inadvertently benefit domestic Chinese competitors like Huawei. He emphasized that China's AI development is deeply intertwined with Nvidia's chips and software ecosystem, suggesting that continued engagement would allow the U.S. to retain some influence over China's technological trajectory. Nvidia also maintains a research center in Shanghai. While Huang's argument has merit, it reveals the tension between corporate interests and national security. Nvidia's dependence on the Chinese market and talent pool may compromise its ability to align with U.S. strategic interests. Investors may view all of this as a slippery slope, in which political considerations begin overriding free-market principles, potentially undermining confidence in U.S. financial markets. Ultimately, the reversal of export restrictions on Nvidia and AMD's AI chips to China reflects a transactional approach to national security—one that prioritizes revenue over resilience. About the author: Christopher Tang is a distinguished professor at the UCLA Anderson School of Management. Guest commentaries like this one are written by authors outside the Barron's newsroom. They reflect the perspective and opinions of the authors. Submit feedback and commentary pitches to ideas@
Time of India
2 days ago
- Time of India
Seven months after stunning the world, China's DeepSeek AI leans on US technology for critical upgrade
China's Revolutionary DeepSeek Turns to American Hardware for Upgrade- China's much-hyped DeepSeek project, once touted as a breakthrough in homegrown AI independence, has quietly circled back to American hardware after its gamble on Huawei's Ascend chips fell short. Engineers found the Chinese processors too unstable for large-scale training of the company's new R2 model, forcing DeepSeek to rely again on Nvidia GPUs — the very technology U.S. export rules were meant to keep out of Beijing's reach. The move underscores a hard truth: while China can innovate at the software and algorithmic level, its AI future still hinges on semiconductors designed in California. This reliance not only exposes the fragility of China's self-reliance narrative but also raises pressing questions about how far Washington's export curbs can really go in controlling the global AI race. DeepSeek's Huawei gamble falters When DeepSeek announced it would train its next-generation R2 model on Huawei's Ascend GPUs , the move was hailed in Beijing as proof that China could shed its reliance on American semiconductors. The plan didn't last long. By June 2025, engineers inside DeepSeek privately acknowledged that Ascend chips failed to deliver the consistency required for massive-scale training. Sources familiar with the project told the Financial Times (July 2025) that the Ascend processors suffered from unstable performance, weaker interconnect bandwidth, and a lack of mature software tools — all critical weaknesses in an era when model training can consume tens of thousands of GPUs simultaneously. Live Events The result: Huawei's silicon is still being used, but only for inference workloads (running the trained model), while training has quietly shifted back to Nvidia hardware, the very dependency China's AI sector was under political pressure to escape. Why Nvidia still dominates China's AI race Despite U.S. export controls, Nvidia's grip remains unshaken. Even China's most sophisticated AI companies struggle to replicate the CUDA software ecosystem that Nvidia has spent nearly two decades refining. Training a model like DeepSeek's R2 — rumored to involve over 700 billion parameters — is less about raw chip speed and more about orchestration, driver support, and optimization libraries. In practice, Chinese engineers describe Huawei's platform as 'running a marathon in sandals while Nvidia wears carbon-fiber spikes.' That blunt analogy, shared by one engineer who worked on both setups, captures the gap. Hardware is only half the battle; software maturity is the other half, and Nvidia still leads. How DeepSeek got its Nvidia chips Here's the thorny part. U.S. rules technically bar Nvidia from selling its most advanced chips, such as the A100 and H100, directly to China. Yet congressional investigators revealed in April 2025 that DeepSeek somehow amassed tens of thousands of Nvidia GPUs through shell distributors in Singapore and the Middle East. Lawmakers on Capitol Hill now accuse DeepSeek of sidestepping export rules, with several Republicans calling for tighter scrutiny. Nvidia, for its part, points to its H20 line of 'downgraded' GPUs — designed to meet U.S. restrictions — which it can legally export. But multiple industry insiders note that DeepSeek's scale suggests it also tapped into backchannels to acquire restricted units. This dual reality highlights Washington's dilemma: export bans slow China, but they don't stop it. What this means for China's AI ambitions DeepSeek's return to Nvidia exposes the contradiction at the heart of Beijing's technology push. On one hand, China is pouring billions into domestic chip design and production. On the other, its most visible AI breakthrough still leans on American silicon . For policymakers in Washington, this is both reassurance and alarm. Reassurance, because it confirms that U.S. technology remains indispensable. Alarm, because despite layers of export controls, Chinese companies are still finding ways to secure the hardware. For Chinese AI startups, the lesson is more pragmatic: innovation at the algorithmic level — as DeepSeek demonstrated with its mixture-of-experts architecture that slashed training costs — can stretch limited resources, but it cannot fully replace cutting-edge chips. The bigger picture: an AI arms race with supply chain chokepoints The DeepSeek saga isn't just about one company's hardware choices. It underscores a larger geopolitical reality: the global AI race hinges not only on data and talent but on who controls the chip supply chain . China's vulnerability: Without access to advanced lithography tools (still dominated by ASML in the Netherlands), domestic fabs cannot produce chips on par with Nvidia's. U.S. leverage: Washington's chip restrictions remain the single most effective lever in slowing Chinese AI ambitions. Market implications: Nvidia stock surged past a $3 trillion market cap in June 2025 , driven in part by relentless demand from both Western hyperscalers and Chinese firms willing to pay premiums through gray channels. As one Beijing-based venture capitalist told: 'Every AI startup pitch deck begins with the same line — how many Nvidia GPUs they can get. Nothing else matters until that question is answered.' China's DeepSeek may represent cutting-edge algorithmic ingenuity, but its reliance on American hardware reveals the fragile foundation of the country's AI push. For now, the future of Chinese AI still runs, quite literally, on Nvidia. The critical question going forward: can Beijing close the gap before Washington tightens controls further? Or will the world's most ambitious AI firms continue to operate in a paradox — building revolutionary software on hardware they cannot officially buy? Either way, the story of DeepSeek's pivot back to U.S. chips is a reminder that in the AI arms race, semiconductors remain the real battlefield. FAQs: Q1. Why is China's DeepSeek using American Nvidia hardware instead of Huawei chips? Because Huawei's Ascend chips were unstable for large-scale AI training, forcing DeepSeek to return to Nvidia GPUs. Q2. What does DeepSeek's reliance on Nvidia mean for China's AI future? It shows China's AI breakthroughs still depend heavily on U.S. semiconductors despite domestic innovation efforts.
