Nvidia pushes hardware but experts say UK AI needs something else
There's disagreement on what it would take to turn the United Kingdom into an artificial intelligence powerhouse.
Nvidia, the world's largest supplier of graphics processing units (GPUs), has called on the UK to boost hardware investment to catch up with the United States and China in the global AI race.
During his recent visit to the country, Jensen Huang, co-founder and chief executive of Nvidia, told UK Prime Minister Keir Starmer that the UK could become the world's third-largest AI ecosystem.
'The UK has the third-largest AI venture capital (VC) investment in the world. The two largest are the US and China, which is fairly obvious,' Huang said in a panel discussion with Starmer at the London Tech Week on Monday.
'The UK has one of the richest AI communities anywhere on the planet, the deepest thinkers, and the best universities … and you're rich with great computer scientists. It's a fantastic place for venture capital to invest.'
He said the UK is in a 'Goldilocks circumstance' or a 'just right' situation where the country has both investors and scientists to develop AI. (For any reader who doesn't know the children's story 'Goldilocks and the Three Bears,' a young girl named Goldilocks wanders into the home of three bears and finds that one of the bowls of porridge on the table is too hot, another too cold – but the bowl for the small bear is at the right temperature.)
Then Huang pointed out that the UK lacks the hardware infrastructure to create an AI ecosystem that can compete with the US and China.
'If you are a particle physicist, you need a linear accelerator. If you are an astronomer, you need a radio telescope,' he said. 'If you're in the world of AI, you can't do machine learning without a machine.'
Huang said he was among a group of technology entrepreneurs, including Google's former Chief Executive Eric Schmidt, Wayve's Chief Executive Alex Kendall and executives of Synthesia and Elevenlabs, at an event hosted by Starmer on June 8. He said the UK government is committed to AI development.
'We're going to invest in helping start the AI ecosystem' in the UK, he said. 'Infrastructure will enable more research, breakthroughs and companies.… Then the flywheel will start taking off. It's already quite large, but we've just got to get that flywheel going.'
On Monday, on the same stage, Starmer announced an additional £1 billion ($1.3 billion) of funding to boost the country's AI compute power by 20 times. He also announced the government's plan to invest £185 million and partner with 11 companies to train 7.5 million UK workers, one-fifth of the country's workforce, in essential skills to use AI by 2030.
In January this year, the Labor government unveiled the AI Opportunities Action Plan, saying it would set out a long-term plan for the UK's AI infrastructure needs, backed by a 10-year investment commitment. It said it had attracted £39 billion of private investment to the AI sector since it took office in July 2024.
The Joint Academic Data Science Endeavour (JADE) consortium, comprising 20 UK universities and the Turing Institute, uses Nvidia's technologies for its AI development. For example, the University of Manchester uses the Nvidia Earth-2 platform to develop pollution-flow models. The University of Bristol's Isambard-AI supercomputer focuses on climate modeling and next-generation science.
While Huang says insufficient hardware facilities are the main obstacle to forming the UK's AI ecosystem, a research report published by Tech Nation, a unit of the Founders Forum Group, said the problem lies in the lack of growth funds and exit opportunities.
The report said the UK is home to more than 17,000 VC-backed startups. It said UK technology startups have raised over $7 billion in VC investment so far this year, 43% of which originated from funds in the US.
'UK founders rate the UK as a good place to start a tech company, but they are less positive about scaling or exiting their companies in the UK,' the report said, citing its survey on more than a thousand UK technology firms.
It said 43% of UK founders it surveyed are considering relocating their company's headquarters outside the UK.
'Almost all of the founders we surveyed who are considering relocating are targeting the US,' it added. 'Of those, more than one in three are looking for better funding availability, exit opportunities and access to a larger market outside the UK.'
According to the survey, half of the founders surveyed suggested that the UK government provide them with tax credits or use a sovereign wealth fund or a co-investment fund to support the growth of their businesses.
The National Wealth Fund, the UK's sovereign wealth fund, mainly invests in green hydrogen, carbon capture, ports, gigafactories, and green steel sectors.
Balderton Capital, a UK-based venture capital firm, said AI startups in the UK raised $15.9 billion last year, compared with $3.1 billion four years ago.
According to PitchBook Data, AI startups in the US raised a record $97 billion last year. VCs poured $209 billion into US startups in 2024, compared with $61.6 billion in Europe and $75.9 billion in Asia.
It's unclear how the Starmer administration's £1 billion long-term funding can help the UK change the game in the global race.
Meanwhile, China seems to be in another extreme – too much investment but too little experience.
A study by the Stanford Center on China's Economy and Institutions showed that China's government VC funds had actually invested $912 billion in startups in the country from 2013 to 2023, about 23% of which were directed to 1.4 million AI-related firms. That amounted to $209 billion in total, or $150,000 each.
The study said 4,115 AI firms received government and private VC investments from 2000 to 2023. Most of these firms initially received government VC investments and sought private investments.
A report published by the MIT Technology Review in March this year said that China had built hundreds of AI data centers in recent years, but many of them have become 'distressed assets' after many AI projects failed.
Read: US Trojan horse alarms pushing China's robots to Europe
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
South China Morning Post
21 hours ago
- South China Morning Post
Caught in the chip war, Malaysia must rethink its US–China balancing act
Malaysia has landed in another tight spot – this time over claims Chinese engineers may have accessed high-end Nvidia chips on its soil to train artificial intelligence models, potentially breaching US export controls. Advertisement The timing could hardly be worse. Trade Minister Tengku Zafrul Aziz and Second Finance Minister Amir Hamzah Azizan were in Washington this week to negotiate down a steep 24 per cent 'reciprocal tariff' imposed by US President Donald Trump in April. Now, their efforts risk being overshadowed by fears in Washington that Malaysia is serving as a backchannel for Chinese firms to access restricted chips – fears that could harden American attitudes and demands. The report, first published by the Wall Street Journal, claimed that an unnamed Chinese firm had booked out Malaysian data centres equipped with Nvidia's most advanced chips. US policy restricts the sale or export of these chips to China and its military-linked entities, including via third countries. A banner showing Nvidia branding at the AI & Big Data Expo 2025 in London on February 5. Photo: Reuters While it is unclear whether any laws were broken, the perception alone could prove damaging. Analysts expect Washington may use the case as leverage to push Malaysia into stricter enforcement of US export controls, especially given the billions of dollars in investments by American tech firms in Malaysian plants and data centres.
South China Morning Post
2 days ago
- South China Morning Post
Malaysia ‘stands firm' on US chip sanctions after alleged workaround by Chinese company
Trade authorities in Malaysia have launched an investigation into claims that Chinese engineers bypassed US chip curbs by renting data centres packed with high-end Nvidia chips to train their artificial intelligence (AI) model in the country. Advertisement The allegation, made in a report by The Wall Street Journal last Thursday, poses a potential stumbling block to Malaysia's tariff negotiations with the US , which has curbed China 's access to high-end chips, including through third nations. Malaysia's trade ministry said it was in the process of verifying the claims. Though the activity might not necessarily breach local laws, the ministry said it 'stands firm' against attempts to work around export controls or engage in illicit trade activities. 'While Malaysia maintains a neutral position on unilateral sanctions, companies operating here have been advised to adhere to other countries' unilateral export controls … to avoid any secondary sanctions on their businesses,' it said in a statement on Wednesday. 01:38 China a 'key market', says Nvidia CEO Huang during Beijing visit as US bans AI chips China a 'key market', says Nvidia CEO Huang during Beijing visit as US bans AI chips A team of four Chinese engineers allegedly flew into Malaysia from Beijing in March, each carrying a suitcase with 15 hard drives containing 80 terabytes' worth of spreadsheets, images and video clips to train an AI model, according to the New York-based newspaper. Advertisement
South China Morning Post
3 days ago
- South China Morning Post
DeepSeek's updated R1 AI model equals coding ability of Google, Anthropic in new benchmark
The latest model update from Chinese artificial intelligence (AI) start-up DeepSeek has matched the coding performance of industry heavyweights Google and Anthropic, according to the latest results from WebDev Arena, a real-time AI coding competition. The updated version of DeepSeek-R1 tied for first place with Google's Gemini-2.5 and Anthropic's Claude Opus 4 on the WebDev Arena leaderboard, which evaluates large language models (LLMs) on their ability to solve coding tasks quickly and accurately. The Hangzhou-based company's R1 scored 1,408.84, in line with Opus 4's 1,405.51 and Gemini-2.5's 1,433.16. The quality of the models' output is evaluated by humans, who determine the scores. DeepSeek's reasoning model has consistently performed at levels close to leading models in various benchmark tests since it was unveiled in January, despite significantly lower training costs. DeepSeek quietly updated R1 in late May, marking its first revision since its high-profile debut. The start-up released R1-0528 on the open-source AI developer community Hugging Face, calling it a 'minor upgrade' and offering no details on the changes. It later said the updated model had improved in reasoning and creative writing capabilities, with a 50 per cent reduction in hallucinations – instances where AI generates misleading information with little factual basis. The R1 update attracted attention from the developer community amid widespread anticipation for DeepSeek's next-generation reasoning model, R2. The company has said little about when it might release its big follow-up.
