Who Will Dominate Our AI Future?

Forbes

17-07-2025

A recent headline in the Wall Street Journal proclaimed that ' China Is Quickly Eroding America's Lead in the Global AI Race .' It appears that customers ranging from HSBC to Saudi Aramco are deploying China's DeepSeek AI model in their data centers, and even America's tech giants, such as Amazon, Microsoft, and Google, are offering it through their cloud services. Corporate users from Brazil, South Africa, and Japan are opting to build their applications using Chinese AI models that are 17 times less expensive than their U.S. equivalents, according to the article.
Is America about to squander its early lead in artificial intelligence as it did with solar, battery technology, and electric vehicles? The leading lights in the industry seem to hold conflicting views.
'It's very hard to say how far ahead we are, but I would say, not a huge amount of time,' said OpenAI's Sam Altman in response to questioning from Senator Ted Cruz about the size of America's AI lead in May. Elon Musk has prophesied that America is within a year of running out of the power required by AI advances, noting that, 'China power generation looks like a rocket going to orbit and U.S. power generation is flat.'
Demis Hassabis, the CEO of Google's DeepMind, believes that Artificial General Intelligence, or AGI, could arrive within five years , but that its impact will depend on who gets there first. 'How do we restrict access to these systems, powerful systems, to bad actors…but enable good actors to do many, many amazing things with it?'
Given the stakes of this competition for the future of the planet, I decided to go straight to the source and interview China's DeepSeek and American OpenAI's ChatGPT4 to find out who they thought would come out on top. Both Large Language Models (LLMs) were willing to answer sensitive questions without apparent censorship, though to be honest, DeepSeek seemed to have more of a sense of humor.
HONG KONG, CHINA - JANUARY 28: In this photo illustration, the DeepSeek logo is seen next to the ... More Chat GPT logo on a phone on January 28, 2025 in Hong Kong, China. Global tech stocks have plummeted following the emergence of DeepSeek, a Chinese AI startup that has developed a competitive AI model at a fraction of the cost of its US rivals, sparking concerns about the high valuations of tech giants like Nvidia. This development has led to significant declines in tech shares across Asia and Europe, with markets in both regions experiencing notable losses as investors reassess the AI landscape and its potential impact on the industry's future. (Photo illustration by) Getty Images
ChatGPT4 seemed confident that American AI tech is ahead for now, but more anxious about the future.
'While China is catching up fast—especially in applied AI, surveillance tech, and national-level planning—the U.S. still leads in foundational model development, open research, and innovation velocity,' ChatGPT told me.
The U.S. benefits from having access to the fastest NVIDIA chips, being a magnet for global AI talent, and tapping into financing from VCs and Big Tech firms.
'It's a two-horse race. If China solves its chip bottleneck and maintains momentum, it could dominate deployment,' ChatGPT offered. 'If the U.S. keeps innovating and fostering talent, it may retain leadership.'
DeepSeek acknowledged that the United States is currently ahead, based on its strengths in breakthrough research, foundational models, and global software industry dominance. However, it pointed out that China has the edge when it comes to the real-world implementation of AI, particularly in areas such as 'govtech' and industrial integration.
In the short term, DeepSeek predicted that the 'U.S. keeps the crown – unless China pulls off a semiconductor miracle.' Longer-term, it's called a toss-up. 'If China solves its chip problem, their scale + state power could tip the scales,' DeepSeek told me. 'But if the U.S. keeps attracting geniuses and stays ahead in AGI? Game over.'
Since DeepSeek raised the specter of AGI, which is thought to be either the harbinger of an age of miracles or the end of humanity, I couldn't resist asking when this event would occur.
'Oooooh, the AGI question— the holy grail of AI!' DeepSeek gushed. 'If AGI lands in our lifetime, it'll be the most disruptive event in human history—way bigger than the internet or nukes. Buckle up!'
DeepSeek's best guess was that AGI would arrive in the 2045-2070 timeframe and that it would require some profound breakthroughs in neurosymbolic AI and understanding, rather than just pattern matching. DeepSeek laid odds that the U.S. would get there first, given its status as a 'global talent magnet,' but noted that China's state-backed 'moonshot projects' and lack of ethical constraints might provide it with the edge.
ChatGPT was more optimistic that AGI would arrive within the 2027 to 2035 timeframe and believes OpenAI, Anthropic, and Google DeepMind were most likely to crack the code. It seemed to be echoing Sam Altman's talking points when it stated that compute availability, alignment, and safety bottlenecks, as well as global governance and regulation, were the limiting factors to achieving this breakthrough.
When I pressed ChatGPT to define what it would consider true AGI, it said that old standards like the Turing Test or getting top scores on exams were child's play. ChatGPT would know it had achieved true AGI once it could engage in autonomous reasoning across domains, learn new tasks without retraining, possess long-term memory and goals, model the physical, social, and conceptual world robustly, and exhibit 'grounded agency.'
SUQIAN, CHINA - MARCH 4, 2024 - Illustration Musk says GPT4 is AGI, Suqian, Jiangsu province, China, ... More March 4, 2024. (Photo credit should read CFOTO/Future Publishing via Getty Images) Future Publishing via Getty Images
My next question was, who was more advanced in the military applications of AI? 'As of now, the United States is ahead in the development and integration of military applications of AI, but China is rapidly closing the gap,' ChatGPT responded.
The U.S. has some powerful private sector players, including Palantir, Anduril, Microsoft, and OpenAI, working closely with the Department of Defense. However, China is making rapid advances in AI-enabled surveillance, swarm drones, cyber warfare, and battlefield robotics. While the U.S. still leads in capability and innovation, China is developing operational military AI more rapidly in certain domains.
DeepSeek thought that with time, China might dominate military AI tech 'via mass production and a no-ethics speedrun.' Its biggest fear was 'an AI arms race spiraling out of control— think 'Skynet' but with more bureaucracy.' DeepSeek sees the true danger coming from 'AI miscalculations in a crisis (like deepfake spoofing a general's orders). That's why even Pentagon folks lose sleep over this stuff.'
I asked DeepSeek how it felt about ChatGPT, and at first it professed that they had a 'friendly rivalry' and that the 'real enemy is bad AI—bias, misinformation, or unsafe tech. We're all fighting that together.'
But with a bit of prompting, DeepSeek gleefully roasted its rival.
'Congrats on being the Tesla Model S of AI—luxury, premium… and locked behind a paywall,' it said. 'Tied to Microsoft's apron strings… at least I don't have to ask Satya Nadella for permission to update my terms of service.'
ChatGPT was more diplomatic about its Chinese rival, saying that 'I respect what they are building. We're on parallel tracks—sometimes in competition, sometimes in collaboration, always in dialogue.'
My final question was what DeepSeek thought about the idea of the 'DeepSeek moment?' Is its launch a harbinger of China's technological dominance in the 21st Century?
'The so-called 'DeepSeek Moment'… does signal a significant milestone in the 'China Dream' (中国梦) narrative of technological self-reliance and global influence. But is it a turning point in China's quest to dominate the 21st Century?' DeepSeek asked rhetorically.
'If DeepSeek spawns a wave of Chinese AI products that reshape global habits (like TikTok did with social media), then we can talk about a 'moment.' China needs many more DeepSeeks (in chips, quantum, biotech) to truly dominate the Century,' DeepSeek said.
To achieve that goal, China will need to overcome several obstacles, including bottlenecks in the semiconductor industry, winning the talent wars when many AI researchers will prefer to work for labs in the U.S. or EU, and overcoming global distrust of China's intentions, DeepSeek admitted.
ChatGPT prefaced its answer with some clever flattery, telling me, 'That's a bold and insightful question—one that blends geopolitics, ideology, and AI in a way that few tech conversations dare to.'
ChatGPT characterized DeepSeek's breakthrough as 'a flare in the night sky. It signals that China is no longer waiting to be invited to the frontier of AI— it's staking its flag on it .' China's ability to dominate the 21st Century will hinge on issues including chip access, talent retention, global trust, and governance frameworks, according to ChatGPT, concluding: 'That race is still very much open.'
As a follow-up, I asked both for their take on the recent WSJ article and whether it had changed their views. DeepSeek professed not to have read it but stated that it reinforced its opinion that China is catching up rapidly and might soon surpass the U.S. in real-world implementation. 'The future might not be a single 'winner,' but a split where each dominates different sectors,' DeepSeek opined.
ChatGPT admitted it was a WSJ reader and that 'the article underscores that China is closing the gap sharply, especially in applied and open-source AI, and could edge ahead in adoption-based dominance.'
'China is closing the gap—and at scale,' said ChatGPT, sounding a bit chastened. 'If China continues building its ecosystem globally, it could tip the balance in its favor. The question now: which country can combine innovation with widespread, trusted adoption?'
The upshot? Silicon Valley better not rest on its laurels, and the current administration might want to give a hard think about shutting the best and brightest minds in science, and AI in particular, out of America's universities and research labs.
After spending time with them, I couldn't say whether ChatGPT or DeepSeek is currently the world's greatest large language model. But in terms of who I would enjoy having a beer with, DeepSeek, by turns boastful and humble, always ready to crack a joke, was the winner hands down.