Hello, this is Greg Canavan, Investment Director at Fat Tail Investment Research. As I step in for Bill Bonner today, I aim to provide some valuable insights into a topic that's dominating financial discussions: the artificial intelligence boom. While I may not match Bill's eloquence, I hope to offer a clear perspective on the longer-term implications of AI on our economy and financial markets.
It's crucial to consider these dynamics because ignoring them could significantly impact your investment portfolio in the coming years. We're not just talking about the obvious beneficiaries like computer chip manufacturers and data center operators; the real game-changer lies in the second-order effects that AI will unleash across various sectors.
Questions abound: Will widespread AI adoption lead to mass job displacement? What will be the ripple effect on housing markets, consumer spending, and the banking sector? While we don't have definitive answers, one thing is certain: AI is a monumental force, and its impact will be profound. Our team is dedicating considerable thought to these complex issues.
Presently, the market seems to be in a state of pure optimism regarding AI, perceiving only opportunities and no immediate risks. This sentiment is fueled by an enormous influx of capital into the sector, creating a seemingly unstoppable momentum.
The Unprecedented Wave of AI Investment
The scale of investment in AI infrastructure is truly staggering. Leading hyperscalers — Amazon (AWS), Microsoft, Google, and Meta — are projected to collectively spend between US$345 billion and US$350 billion in 2025 on data centers alone. This represents a remarkable year-on-year increase of over 50% compared to 2024. Looking further ahead, their combined capital investment is anticipated to exceed an astounding US$1 trillion between 2025 and 2027. These figures underscore the fierce competition and conviction that these tech giants have in the future of AI.
Independent analyses corroborate this trend. McKinsey, for instance, forecasts that by 2030, nearly US$7 trillion will be allocated to data center infrastructure. This includes substantial investments not only in advanced AI hardware but also in the essential facilities and supporting grid infrastructure required to power this new technological era. It's no wonder that with such colossal capital flows, the market continues its upward trajectory, seemingly immune to broader economic concerns.
The Looming Energy Challenge and Grid Constraints
However, beneath this veneer of limitless growth lies a critical constraint that garnered significant discussion within our team: energy. The sheer energy demands of this burgeoning AI infrastructure pose a substantial challenge to global data center rollout plans. A recent report by the International Energy Agency (IEA), titled ‘Energy and AI’ (released in April), highlights this growing concern.
The IEA projects a dramatic increase in global data center electricity consumption, from 415 TWh per year in 2024 to a staggering 945 TWh per year by 2030. This massive surge in energy demand over a relatively short period presents an unprecedented test for global energy grids. The United States is expected to account for the largest share of this increase, closely followed by China. Together, these two nations are projected to represent nearly 80% of the global growth in data center energy consumption. This makes perfect sense given their intense geopolitical race for AI supremacy, which naturally prioritizes securing low-cost energy.
In stark contrast, countries like Australia, with high energy costs and a fervent commitment to renewables (as championed by figures like Chris Bowen), may find themselves at a disadvantage, potentially becoming an ‘AI backwater.’ The reality is that the age of AI will necessitate all forms of energy. Renewables, battery storage, coal, gas, and even nuclear power will all need to expand significantly over the next five years to adequately support the global AI rollout. US real estate company JLL reports approximately 10 GW of new data center capacity currently under construction worldwide—an electricity demand equivalent to powering several medium-sized nations. It is a reasonable assumption that this relentless AI race will soon encounter serious electricity grid constraints, as there is a practical limit to how much new capacity can be integrated into the grid, and how quickly.
Questioning the Return on Investment
Beyond energy, another fundamental question will inevitably arise: What is the return on investment (ROI) for each incremental dollar poured into AI? Initially, the returns on AI investments appear highly attractive, primarily because they enable hyperscalers to drastically improve efficiency and reduce their workforce. Reports indicate that these efficiency gains could lead to a loss of 235,000 tech jobs globally in 2025 alone. This immediate benefit incentivizes further investment.
However, the tech sector can only absorb so many job losses before the wellspring of efficiency gains begins to run dry. At some point, companies will need to identify new avenues to justify continued incremental investment in AI. This is the critical juncture when the market, currently captivated by opportunity, might start asking difficult questions about profitability and sustainability. The fundamental issue is that we are in an AI arms race; every major player is building capacity, driven by the fear of being left behind. The consequence of this ‘build-at-all-costs’ mentality, regardless of immediate need, is a looming threat of overcapacity, which inherently leads to declining returns on investment.
Predicting the exact timing of this market recalibration is challenging. If I had to venture a guess, I would speculate that by mid-2026, we might start to see these pressures emerge. But it remains just that—a guess. Furthermore, consider what happens when the profound efficiency gains currently enjoyed by the tech sector permeate every other industry globally. While the AI boom will undoubtedly create new jobs, it's highly probable that the initial net effect on employment will be negative. In Australia, given its existing productivity crisis and the limitations imposed by high energy costs on data center development, a negative net impact on the local job market seems a solid bet. This scenario will, of course, create winners and losers among companies.
The Old Economy as a Stabilizer and the Looming Bust
On a positive note for Australia, AI cannot magically produce iron ore, gold, coal, or liquefied natural gas (LNG). The world will continue to demand these essential commodities. In this context, the 'old economy'—resource extraction and production—might once again prove to be a crucial stabilizer for our national economy. So, while the AI-fueled boom continues, with more fuel in its tank, it's wise to remember that as more companies adopt AI, the rate of job losses will likely escalate.
This rising unemployment could have significant ramifications for economies heavily reliant on debt and high house prices. While this might appear to be a problem for 2026 and beyond, it is a critical consideration that thoughtful investors and policymakers should be addressing now. The journey through the AI revolution promises unprecedented innovation, but it also demands a sober assessment of its potential pitfalls and the eventual bust that often follows a boom.
