The ultimate GenAI showdown: US v China
On GenAI, the US is ahead - but neither Chinese Tech nor Beijing want to be left behind
Happy Friday All,
I’m coming to the end of a three-month AI speaking tour — and from every corner of the globe, the AI revolution is not just afoot; it's hurtling forward at a breakneck pace.
This is playing out at the highest level of geopolitics. And the ultimate GenAI showdown between the US and China is starting to gather pace and speed.
‘Capitalism at its best’
Right now, the US is still comfortably ahead.
Part of this story has to do with money. When Open AI was founded with a $1bn founding endowment in 2015 — profit wasn’t the main concern. Instead, it was about the advancement of intelligence for humanity.
Such lofty founding principles, combined with an abundance of private money to power innovation, have put the US streets ahead in GenAI. According to the Stanford AI Index, global private investment in AI in 2022 was $91.9 billion — the bulk of which came from America.
This week, we had news of another huge raise for a US AI company. Inflection AI, the brainchild of tech titan Reid Hoffman and Mustafa Suleyman (formerly of DeepMind), raised $1.3 billion for ‘Pi,’ their conversational chatbot that was launched just two months ago. Notable backers include Microsoft and Nvidia, as well as Bill Gates and Eric Schmidt.
And what’s a cool ‘bill’ in LLM development these days anyway? Antrophic, a company backed by Google, and co-founded by OpenAI alumni to compete with OpenAI, had also raised almost a billion dollars by May of this year.
However, while the AI funding pool has increased significantly over the past decade, the number of companies receiving these funds has actually dropped. Instead, we are starting to see fewer companies receiving larger investment rounds, signalling the concentration of power and capital in the AI space. (i.e., Microsoft’s $10bn investment in OpenAI.)
This is capitalism at its best. You want capital to chase opportunity and that drives creativity and invention…Of course, some people are going to lose their shirts.
Mustafa Suleyman, Inflection Co-Founder
China’s GenAI Strategy?
So what’s happening in China?
According to previously unreported data collated by consultancy Preqin, AI investment in China in the year to mid-June stands at $4bn. This is dwarfed by US investment for the same period, which stands at $26.6bn.
Interestingly, all the major innovations in the new wave of GenAI have come from the ‘West’ — not from China. Nonetheless, in the post-ChatGPT world, GenAI is emerging as a new obsession for Chinese tech titans, who do not want to miss out on the ‘iPhone moment’ for this generation.
In the frenzied few months since ChatGPT’s release, Chinese contenders from Baidu to Alibaba have all launched their own versions to challenge OpenAI’s ChatBot. (Baidu recently announced that its ChatGPT-style service outperformed OpenAI's original product on several metrics.)
Even the administration in Beijing, which has a ‘complicated’ relationship with Chinese tech billionaires, understands that leaning into GenAI will be critically important for securing China’s ascendancy on the global stage.
Beyond the Numbers: AI's Deep Implications
Meanwhile, it has long been in Washington’s strategic interest to deter China’s AI ascendancy. The Biden administration is accelerating its efforts to curb China’s ability to get ahead in GenAI.
Last October (pre-ChatGPT!) the US moved to block exports to China of the critical hardware needed to power the AI revolution: chips. Top chips, including Nvidia's A100 — and the chipmaking tools needed were restricted as exports to China.
(This is against a backdrop of a general ‘chip shortage.’ In the words of Elon Musk, they are harder to buy ‘than drugs.’)
Now, the WSJ reports that the Biden administration is thinking of further curtailing Chinese companies’ abilities to access US hardware. As the regulations currently stand, Chinese companies can also effectively lease access to top-of-the-line chips from cloud computing centres overseas, however, the US government is considering restricting that access from as early as next month.
This would not only have implications on the AI race, but also for the cloud services space, currently dominated by the likes of Amazon and Microsoft.
While Beijing is intent on designing and manufacturing its own chips and thus weaning itself off its dependence on American hardware in the long term, the Biden administration's bet is that regulation and investment will keep the US ahead in the AI arms race.
For now, it is.
But this is only the opening act in what is going to become a long and protracted geopolitical struggle. Let’s watch this space.
LLMs are math & bytes; export restrictions will not be practical in the nuclear non-proliforation sense. The bleeding edge is only 6 months to a year ahead of the open source / commodity hardware versions.
It's more about the large scale projects these tools are applied to being grounded in ideologies of western democracies vs not.
Would be interesting to do interviews and better understand what applications tech innovators in the east see applicable. Are they working to apply LLMs to scale government social ranking systems, build in a particular world view in censorship/alignment of the systems that will replace search engines, news aggregation and social network rank systems ?
How can we apply usage of the tools in western democracies to push towards liberals notions of decentralized control; scientific reasoning. We may be failing here already with heavy "alignment" to particularly ideologies in the centralized mainstream LLMs that most western individuals are interacting with ?