The rise of DeepSeek’s AI models is offering Chinese chipmakers like Huawei a better chance to compete in the domestic market against the powerful processors from US companies like Nvidia. For years, Huawei and other Chinese manufacturers have struggled to match Nvidia’s high-end chips, which are essential for training AI models. However, DeepSeek’s focus on ‘inference’ rather than raw processing power has led analysts to believe that it could help close the gap between Chinese-made processors and their US counterparts. Inference refers to the phase where AI models use trained data to make predictions, a process less reliant on heavy computational resources.
Several Chinese AI chipmakers, including Huawei, EnFlame, and Moore Threads, have recently stated that their products will support DeepSeek models, although few details have been disclosed. Industry executives predict that DeepSeek’s open-source nature and its low fees will drive the adoption of AI, helping Chinese companies bypass US export restrictions on advanced chips. In fact, Chinese chips like Huawei’s Ascend 910B have already been recognised as better suited for inference tasks, which require less computational power than training.
Despite these developments, Nvidia still dominates the global AI chip market. Analysts point out that while Chinese chips are cost-effective for inference tasks, Nvidia’s superior chips remain the preferred choice even for inference. Nvidia’s CUDA platform, which provides developers with a robust software environment, remains a key advantage, and Chinese companies like Huawei have struggled to convince developers to abandon CUDA in favour of their platforms, such as Huawei’s Compute Architecture for Neural Networks (CANN). The software performance of Chinese AI chips continues to lag behind, making it challenging for them to directly challenge Nvidia’s dominance.
For more information on these topics, visit diplomacy.edu.
China’s spending on chipmaking equipment is expected to fall by 6% in 2025, marking its first decline since 2021, according to Canadian research firm TechInsights. The drop follows years of aggressive stockpiling as Chinese firms sought to bypass tightening US export controls. Last year, China accounted for 40% of global semiconductor equipment purchases, but its share is projected to shrink significantly.
Analysts attribute the decline to a combination of overcapacity and the impact of US sanctions, which aim to limit China’s ability to develop advanced chips with potential military applications. Despite these restrictions, companies like SMIC and Huawei have continued to push forward, achieving technological breakthroughs by relying on more expensive and labour-intensive manufacturing processes.
China has made strides in producing mature-node chips and expanding its domestic semiconductor industry, with firms like Naura Technology Group gaining global market share. However, the country remains dependent on foreign suppliers for critical tools such as lithography machines, highlighting ongoing challenges in its push for self-sufficiency.
For more information on these topics, visit diplomacy.edu.
Dario Amodei, CEO of AI firm Anthropic, has warned that the race to develop AI is moving faster than efforts to fully understand it. Speaking at an event in Paris, he stressed the need for deeper research into AI models, describing it as a race between expanding capabilities and improving transparency. ‘We can’t slow down development, but our understanding must match our ability to build,’ he said.
Amodei rejected the notion that AI safety measures hinder progress, arguing instead that they help refine and improve models. He pointed to earlier discussions at the UK’s Bletchley Summit, where risk assessment strategies were introduced, and insisted they had not slowed technological growth. ‘Better testing and measurement actually lead to better models,’ he said.
The Anthropic CEO also discussed the evolving AI market, including competition from Chinese firm DeepSeek, whose claims of dramatically lower training costs he dismissed as ‘not based on facts.’ Looking ahead, he hinted at upcoming improvements in AI reasoning, with a focus on creating more seamless transitions between different types of models. He remains optimistic, predicting that AI will drive innovation across industries, from healthcare to finance and energy.
For more information on these topics, visit diplomacy.edu.
Adobe has launched the first public version of its AI-powered video generation tool, Firefly Video Model, introducing competition to OpenAI’s Sora and Runway’s video-generation services. The tool is designed to integrate with Adobe’s Premiere Pro software, making it useful for film and television professionals. Instead of focusing on generating long video clips, Adobe’s model helps improve or extend real production shots that need adjustments.
The tool currently produces five-second clips at 1080p resolution, shorter than OpenAI’s 20-second limit, but Adobe argues that most production clips are only a few seconds long. Pricing starts at $9.99 for 20 clips per month and $29.99 for 70 clips, with a separate ‘Premium’ plan for high-volume users like studios to be announced later this year. Adobe is also working on 4K video generation, prioritising visual quality over longer clips.
Vice President of Generative AI Alexandru Costin emphasised that Adobe aims to make AI-generated video look as realistic as traditional filming. The company remains focused on improving motion, structure, and image quality rather than extending clip duration. Meta Platforms is also developing a video-generation model, but has not yet confirmed a release date.
For more information on these topics, visit diplomacy.edu.
Germany‘s Federal Cartel Office has expressed concerns over Apple’s App Tracking Transparency (ATT) feature, which could potentially violate antitrust rules for large tech companies. The regulator’s preliminary findings come after a detailed three-year investigation into the feature, which allows iPhone users to block advertisers from tracking their activities across multiple apps. The investigation is part of broader scrutiny over the influence of major tech companies on the digital advertising ecosystem.
In a statement released on Thursday, the Federal Cartel Office noted that Apple now has the opportunity to respond to the allegations. The authority’s concerns focus on whether ATT unfairly impacts the business models of other companies that rely on data-driven advertising, such as Meta Platforms, app developers, and startups. These businesses argue that the feature could severely limit their ability to target users with personalised ads, affecting their revenue generation strategies.
Apple has defended ATT as a crucial privacy tool that empowers users to have more control over their data. The company argues that the feature helps to protect user privacy by giving individuals the option to block third-party tracking. However, its critics, particularly in the advertising industry, contend that ATT has created an uneven playing field, disadvantaging businesses that depend on targeted advertising. The outcome of this investigation could have significant implications for Apple’s business practices in Europe.
For more information on these topics, visit diplomacy.edu.
Thomson Reuters has won a legal battle against Ross Intelligence, after a judge ruled that the law firm’s use of Thomson Reuters’ legal content to train an AI model violated US copyright laws. The case stems from a 2020 lawsuit where Thomson Reuters accused the now-defunct legal research firm of using its Westlaw platform to build a competing AI system without permission.
Judge Stephanos Bibas confirmed that Ross Intelligence’s use of the content did not qualify as “fair use” under US copyright law, which permits limited use of copyrighted material for purposes such as teaching or research. Thomson Reuters expressed satisfaction with the ruling, stating that copying its content for AI training was not a fair use.
This case is part of a broader trend of legal challenges involving AI and copyright issues, with authors, artists, and music labels filing similar lawsuits against AI developers for using their works without compensation. These cases all involve the claim that tech companies have used vast amounts of human-created content to train AI models, raising concerns about intellectual property rights and the ethics of AI development.
For more information on these topics, visit diplomacy.edu.
The Dutch government plans to cut bureaucracy and increase investment in artificial intelligence to help the country’s tech startups thrive, Prime Minister Dick Schoof announced. His comments follow a report by TechLeap, which revealed a sharp decline in small firms securing significant funding. Despite a 47% increase in venture capital investment in 2024, the number of Dutch startups receiving more than €100,000 fell dramatically, with most funding coming from foreign investors.
Schoof, speaking at TechLeap’s annual event in The Hague, stressed the urgency of creating a business-friendly environment to attract venture capital. He warned that Europe risks being left behind by the US and China if immediate action is not taken. Eindhoven, home to chip giant ASML, has been a key driver of the Dutch economy, but the slowdown in startup growth raises concerns about long-term innovation.
Two Dutch firms, hotel software developer Mews and AI-powered auditing company DataSnipper, achieved unicorn status last year, but industry leaders remain cautious. With the government now vowing to intervene, the hope is that streamlined regulations and targeted investments will help revive the country’s startup ecosystem.
For more information on these topics, visit diplomacy.edu
Google has announced a new partnership with Poland, signing a memorandum of understanding aimed at accelerating the adoption of AI in the country. The agreement was revealed by CEO Sundar Pichai during a news conference on Thursday. Through this collaboration, Google and Poland seek to use AI as a tool to drive economic growth and innovation, focusing on areas that include cybersecurity, health, and energy.
The deal highlights the potential for AI to transform key sectors of the economy, with both parties eager to explore new opportunities. Pichai emphasised how AI could help bolster cybersecurity measures, improve healthcare delivery, and increase energy efficiency in Poland, which would have far-reaching implications not just for the country but for the broader European region. These sectors, in particular, are seen as critical for future development, and the use of AI could significantly enhance Poland’s technological capabilities.
This partnership also underscores Google’s growing commitment to global AI initiatives, as the company continues to collaborate with governments and organisations around the world to implement AI-driven solutions. Poland is positioning itself as an important player in Europe’s AI landscape, and this agreement represents a step forward in its efforts to modernise and compete on a global scale in the rapidly evolving tech space.
For more information on these topics, visit diplomacy.edu.
Reddit is set to overhaul its search functionality in 2025 by integrating AI-powered features to help users find better answers to complex and subjective questions. CEO Steve Huffman announced that the company will incorporate Reddit Answers, a tool that summarises discussions and highlights key responses, into its existing search system.
The company is also forming a dedicated team of engineers to refine and expand these AI-driven search capabilities. This move follows Reddit’s broader AI strategy, which has already introduced translation tools and AI-powered insights for advertisers. The platform has also experimented with AI-curated search result pages to improve content discovery across different communities.
Despite these advancements, Reddit’s latest earnings report disappointed investors, with daily active users reaching 101.7 million falling short of expectations. Huffman suggested that making search an integral part of user onboarding could help boost engagement and monetisation, positioning AI-powered search as a key driver of Reddit’s future growth.
For more information on these topics, visit diplomacy.edu
Elon Musk announced on Thursday that his AI chatbot, Grok 3, is nearing completion and will be released in the next week or two. In a video call at the World Governments Summit in Dubai, Musk stated that Grok 3 exhibits powerful reasoning capabilities, outperforming other AI models in tests conducted so far. This new AI is being developed through Musk’s company, xAI, which he founded to rival OpenAI and Google.
Musk, also a co-founder of OpenAI, has been outspoken in his criticism of the AI startup. Recently, a consortium led by Musk made a $97.4 billion offer to acquire OpenAI’s nonprofit assets, a move that comes amid Musk’s ongoing legal battle with OpenAI. He is challenging the company’s decision to transition to a for-profit model, arguing that it undermines its original nonprofit mission.
In addition to discussing AI, Musk shared his thoughts on reducing US government spending, suggesting that $1 trillion could be saved by cutting inefficiencies. He also spoke on international matters, advising the US to be less involved in other countries’ affairs. Musk’s comments followed a partnership announcement with UAE AI Minister Omar Al Olama on a futuristic underground transport system called the “Dubai Loop.”
Musk’s remarks at the summit highlighted his wide-ranging influence, not only in technology but also in global political and economic discussions.
For more information on these topics, visit diplomacy.edu.
