Fractures in the Global Tech Leaderboard
Will Challenger Countries Leap Frog Ahead in the Age of AI?
(I generated a podcast discussion of the post I wrote below. Click here to listen to the discussion)
Last week the FT ran a controversial piece that discussed the decline of venture backed startups in China.
Critics of the report (including the founder and CEO of the data provider) point out that the report was misleading or that the underlying dataset was incomplete. Controversy surrounding the accuracy of the report aside, I actually think the discussion sparked by the report misses a more important point: looking at historical data is irrelevant, if not outright misleading, when faced with a coming platform disruption, akin to looking into the rearview mirror while driving on a windy road.
The generative AI wave provides an opportunity to re-rank technology prowess for all countries. For better or worse, being a leader under the existing technology paradigm is less relevant when facing a new wave of technology. In 2007, for example, Nokia and RIM were the clear leaders in mobile, but upstarts iOS and Android easily displaced the incumbents with the advent of smartphones. While China may have taken their foot off the throttle on startups in general, the more important question for all countries is how each has grasped the opportunity in the generative AI Wave.
The innovator’s dilemma impact on companies is well-known, but less often discussed is the effect on countries. Have we seen technology shifts cause a re-ranking of countries before? During the 19th century transition from what economist Carlota Perez calls the Second Technological Age (Age of Steam and Railways) to the Third Technological Age (Age of Steel, Electricity and Heavy Engineering), the reigning world superpower Britain ceded economic might to challengers US and Germany. The shift from coal-based fuel to oil gave economic rise to places like Russia, Norway and the Middle East. Generative AI is creating a smaller scale rift that could have major long term implications. Let’s take a look at how the latest technology wave may affect the leaderboard.
Americas
There is no question that US continues its lead in AI. To date, all five frontier model companies that have raised at least $1B are based in the US: OpenAI, Anthropic, Inflection, XAI, and SSI (Canada’s Cohere came close at $970M). The Magnificent Seven has replaced FAANG as the new moniker for leading technology companies, and the cohort remains entirely US-based. The US continues to be the top destination for the most elite AI talent in the world. One vital dependency for LLM supremacy, NVIDIA, is under the jurisdiction of the US. American dominance in AI seems unassailable, for now.
What are the potential threats to US leadership in AI? Policies at the federal and state level, as well as antitrust regulatory enforcement, however well-intentioned, can slow down progress. There is an open question whether copyright laws prohibit model companies from using copyright content in training. US immigration policy will need to keep pace with efforts in Canada, EU and the UK to attract the best AI talent from around the world. Although NVIDIA is in the US, a critical part of the supply chain remains in Taiwan with TSMC (although early efforts reached a milestone as the Arizona fab is making Apple’s A16 chip). As energy becomes a critical bottleneck to scaling data centers, the higher cost of building alternative energy sources in the US can limit a critical part of the supply chain.
May the FT piece be a poignant reminder: building a flourishing startup ecosystem is incredibly difficult and can collapse if not properly managed. Silicon Valley, for instance, continues to be the envy of the world, and its beneficiaries would do well to preserve the unique ecosystem for participants involved.
Europe
The UK has always been the AI lead in the European region, if not the world for a period of time. Leading institutions like Oxford, Cambridge, and UCL, along with companies like DeepMind continue to create a vibrant ecosystem producing AI talent. But just as the Turing Test, which originated from Alan Turing’s 1950 paper, has been surpassed as a benchmark for intelligence, the UK faces intense competition as the European AI champion.
France is the leading contender vying for its leap frog moment during the generative AI wave. France has a long tradition of math education, but more recently in engineering and computer science. The large tech companies have further nurtured talent with AI research hubs by DeepMind, Meta and Microsoft. French founders have claimed important parts of the AI ecosystem in the US with companies like Hugging Face and Snowflake.
So far, these characteristics do not meaningfully distinguish France from the UK. But here’s where France begins to differ from its northern neighbour: the French government threw its weight behind Mistral AI by having its public bank Bpifrance’s invest in Mistral’s $113M seed round, the largest seed round in Europe for an AI company at the time. H (formerly Holistic AI) raised a seed round of $220M and also backed by Bpifrance. So pervasive was the French government support that I saw something I’ve never seen before: an AI company based in SF moved from the undisputed center of the AI universe and relocated to France. Not surprisingly, Poolside AI, the code gen company, also counts Bpifrance as one of their investors and is partnering with Paris-based Scaleway for cloud infrastructure. The size of these rounds are outliers for the seed stage not only in France or Europe, but in the whole world.
In addition to funding, the French government has identified key dependencies for AI supremacy and managed to attract large investments in cloud infrastructure, skills training and startup support. France famously derives up to 70% of its energy needs from nuclear and is in a good position to expand energy production. In addition to protecting its energy policy against critics in the EU, France also pushed back on the EU during the drafting of the EU AI Law by warning against stifling innovation and hampering its ability to compete against the US. President Macron’s administration has identified the opportunity to separate itself from the rest of the EU and leap ahead in the AI race.
Asia
Asia has a similar dynamic of having a regional champion, but unlike Europe, there are no signs that China will give any ground as the AI incumbent in Asia. Despite export controls on the most powerful GPUs to China, Chinese models regularly appear on leaderboards next to SOTA models from the US and France. The competitive pricing from cloud providers in China suggests that NVIDIA’s high end H100’s and A100’s are being smuggled successfully into the country.
“Four small-scale Chinese cloud providers charge local tech groups roughly $6 an hour to use a server with eight Nvidia A100 processors in a base configuration, companies and customers told the Financial Times. Small cloud vendors in the US charge about $10 an hour for the same set-up.“ Source: FT
As for talent, China appears to maintain its position as the second most popular country for elite AI talent. China’s energy policy will be a strong asset as the country expands capacity to support the growing needs of large data centers. In the area of nuclear, China can build new plants for about $2,500 to $3,000 per kilowatt, which is about a third the cost in the U.S. and France.
“Indeed, China has embarked on a rapid buildout of its nuclear industry, with 27 nuclear reactors under construction (more than two and a half times more than any other country) complementing the existing fleet of 56. The country expects to build 6 to 8 new nuclear power plants each year for the foreseeable future, with the country surpassing the United States in nuclear-generated electricity by 2030. In total, China intends to build a total of 150 new nuclear reactors between 2020 and 2035.” Source: itif.org
If you look closely, however, fissures are beginning to form. According to the FT, risk capital has shrunk, so founders will need to appeal to investors prioritising short term, predictable returns. Given how competitive other countries are trying to attract elite AI researchers, retaining talent and preventing a brain drain will be an important focus, not to mention attracting foreign AI talent to China. But a bigger challenge for China will be very specific to where generative AI technology is headed, to be discussed in a separate piece.
If China stumbles during this technology wave, Japan is poised to take the lead in Asia. Prime Minister Kishida has explicitly stated the goal in his Startup Development Five-year Plan to make “Japan the largest startup hub in Asia.” I briefly summarized some of the advantages that Japan has in the AI field as part of the background into 500 Global’s investment in the leading AI research lab Sakana AI:
“While starting in the 1950s at the dawn of the modern AI era, Japan accelerated its support of AI research in the early 1980s with deep collaboration among academia, the private sector, and government support. Researchers in Japan were exploring massively parallel computing for more than a decade before NVIDIA found a commercial market by rendering graphics for PC games. Japan’s expertise in electronics and robotics, along with its current focus on developing healthcare AI, autonomous vehicles, and smart cities, offers an extremely attractive environment for state-of-the-art AI solutions developed domestically.” Source: 500.co
Announced in 2022, the Japan’s Five-year Plan has created a sense of urgency that I have not seen in other governments. Within the last nine months, Japan has attracted large infrastructure investments from hyperscalers Azure with $2.9B, AWS with $15B and Oracle with $8B. I witnessed the speed and urgency first hand as one of the AI companies that I invested in went from seed to unicorn in less than 7 months, the fastest to have done so in Japan. In the process, they received a grant from the Japanese government to access compute, which directly contributed to one of the explicit goals of Japan to create 100 unicorns and 100,000 startups. I’ve worked in technology with various components in Japan over the last 20 years, and the speed of execution by the Japanese government now is formidable.
In addition to nurturing talent at home, Japan has also become more aggressive about attracting foreign talent. To encourage more early stage risk capital and expertise, Japan announced a path towards five-year residency for angel investors. Japan has already relaxed residency requirements for foreign entrepreneurs in an attempt to attract more founders to Japan. This is a key differentiator from China where foreign capital and talent recently left the startup ecosystem. Specific to attracting AI developers, some have interpreted recent moves to suggest that Japan will take Israel’s position of allowing models to train on copyright data, a position more permissive to model developers than even in the US (but the issue is not settled).
Playbook for the Age of AI
There is a much longer list of countries that are making major strides in the age of AI while not contending for outright lead (yet). These include Israel, UAE, and Singapore, and each are approaching the opportunity in unique ways that draw upon the strengths of their home country. These countries fighting to stay ahead, or challenging for the lead, are on the vanguard of developing the new playbook for technology leadership in the age of generative AI. The new platform provides a new generation of founders, those founders have different needs, and countries are taking advantage of the disruption to re-order the global technology leaderboard.