In the autumn of 2026, a delegation from Australia made its way to San Francisco, not for the typical tech tour of startups and venture capital, but on a mission to understand the state of artificial intelligence. What they found was a landscape of breathtaking innovation and profound anxiety, a place where the future of humanity is being written in code. The United States, they learned, is not just participating in the AI revolution—it is leading it, several months ahead of China in what has become a technological sprint with global implications. Yet this leadership comes with a paradox: America is simultaneously the epicenter of AI advancement and the 'wild west of regulation,' a place where powerful systems are developed with few guardrails.
San Francisco, the home of OpenAI and Anthropic, presents a stark contrast of perspectives on artificial intelligence. In one corner, optimists speak of a coming golden age, where AI will solve humanity's most intractable problems, from disease to climate change. In the other, doomsayers warn of existential threats, suggesting that uncontrolled superintelligence could pose an unprecedented risk to human survival. The Australian delegation heard both viewpoints, but it was the sense of urgency that resonated most deeply. This is not a theoretical debate being played out in academic journals; it is a conversation happening in boardrooms and research labs where decisions are being made that could shape centuries to come.
Perhaps the most striking moment came when an AI safety expert offered a chilling piece of advice to the Australians: 'If you have things you want to do in your life, do them by 2027.' The comment, delivered without elaboration but with unmistakable gravity, encapsulated the growing concern among some researchers that the timeline for addressing AI risks is alarmingly short. The pessimists in the field suggest there is a 10-50 per cent chance that humanity could be wiped out as a species in the next couple of years if proper safeguards are not put in place. These are not abstract numbers but the considered estimates of experts who spend their days contemplating the potential consequences of the technology they are helping to create.
As the Australian delegation toured the facilities where cutting-edge AI is being developed, they were struck by the scale and secrecy of the operation.
As the Australian delegation toured the facilities where cutting-edge AI is being developed, they were struck by the scale and secrecy of the operation. One participant observed that there are 'multiple Manhattan Projects being run by private companies'—referring to the massive, clandestine efforts that produced the atomic bomb during World War II. The comparison is both apt and unsettling. Just as the Manhattan Project represented a technological leap with world-altering consequences, today's AI development represents a similar inflection point. The difference is that the race is no longer between nations but between private corporations, each competing to build more powerful systems without always understanding the full implications of their work.
Australia's interest in this American AI boom is not merely academic. As a middle power with significant technological capacity, Australia finds itself at a crossroads—determined to benefit from AI's potential while recognizing the need to mitigate its dangers. The visit to San Francisco was part of the Australian Council on AI Strategy's effort to develop a national approach to artificial intelligence that balances innovation with safety. The Americans' willingness to share insights, even as they acknowledge their regulatory shortcomings, offers valuable lessons for a country still formulating its own AI strategy.
As 2026 draws to a close, the American AI landscape remains a study in contrasts. It is a place of unparalleled innovation and profound uncertainty, where the brightest minds are building systems that may one day surpass human intelligence. The Australian delegation returned home with a clearer understanding of the challenges ahead, carrying with them the sobering warning about the approaching 2027 deadline. The question is no longer whether artificial intelligence will transform our world, but whether we can harness its power while steering clear of the existential risks it may pose. The race is on, and the finish line may be closer than any of us realize.
AMD $AMD has two major product releases converging at the same time: Venice CPUs and the highly anticipated Helios MI400 Series. Just three months ago, management expected server CPU revenue to grow more than 70% in 2027, while total Data Center revenue would grow “well over 100%,” driven by the Helios ramp. Since then, AMD has raised its outlook considerably, and the two product ramps are now exceeding our original expectations. Server CPU revenue is now expected to more than double in 2027, up from the previous outlook for 70%+ growth. Meanwhile, Data Center AI revenue (GPUs) is expected to significantly more than double. All-in, AMD guided for companywide revenue to increase more than 80% next year, which is more than twice the company’s long-term growth model outlined at its November 2025 Analyst Day. Perhaps the most important milestone is that AMD’s Data Center AI business is expected to surpass its server CPU business in revenue for the first time in Q4, as Helios enters a multi-gigawatt cycle with Anthropic, $META, $MSFT, OpenAI, and $ORCL deploying its new racks. This brings us back to my July 2024 prediction that AMD’s returns would surpass Nvidia $NVDA, a call I made when Nvidia was the market darling. Since then, AMD has gained more than 250% compared to Nvidia’s 51%, outperforming by about 5X. And I have a hunch that gap could get wider. Full analysis below. https://t.co/F9vxaE2eKC