The global economy is, by most near-term measures, in reasonable health. Corporate profits are rising, AI-driven investment is booming, and labour markets, while cooling, remain resilient. Yet beneath this constructive surface lies a set of structural forces so profound, and so genuinely uncertain in their ultimate direction, that investors who build portfolios around a single long-term conviction may be making a costly mistake. Fed Chair Kevin Warsh, whatever one thinks of his near-term monetary policy instincts, is right about one thing: this is not a moment for high conviction on the long-run outlook.

The AI enigma

Artificial intelligence sits at the heart of today’s economic narrative – and its contradictions. Tech investment has reached 4.9% of US GDP in Q1 2026, eclipsing even the peak of the dot-com era. Hyperscaler capital expenditure is projected to reach more than USD1 trillion by 2027. The bulls argue AI will be as transformative as the Industrial Revolution, potentially doubling global income within a generation. The bears point out that, for now, AI is raising inflation – pushing up electricity prices, memory chip costs, and energy infrastructure demand – rather than delivering the disinflationary dividend its proponents promise.

The honest answer is that nobody knows. Micro studies suggest AI can make tasks more efficient, but with only a fraction of tasks exposed and adoption still shallow, the aggregate productivity boost may be limited in the next few years. Whether those productivity gains flow to workers or to capital owners will determine whether AI proves to be a broadly shared prosperity engine or a force that deepens inequality and suppresses consumer demand. The long-term impact on inflation and interest rates remains, less clear. This contrasts with the certainty with which AI’s economic effects are often proclaimed.

The debt Supercycle and fiscal fragility

Compounding the AI uncertainty is the uncomfortable reality of government balance sheets. Global debt rose for a fifth consecutive quarter in Q1 2026, hitting a record high of over USD350 trillion. US public debt as a share of GDP is close to its 1945 wartime peak. Interest payments have risen sharply as a share of GDP since 2019, absorbing fiscal space and increasing bond market sensitivity to any inflation surprise.

The world is simultaneously attempting to finance AI infrastructure, defence modernisation, energy security, healthcare for aging populations, and supply-chain resilience. Many call it the emerging Supercycle. Yet, these are not optional expenditures. They are structural. And they compete for a finite pool of global savings, pushing desired investment above desired saving and keeping real interest rates structurally elevated. The old low-rate world, built on weak investment, abundant labour, efficient supply chains, and subdued inflation, is gone.

Deglobalisation, demographics, and the labour supply squeeze

The disinflationary tailwinds that defined the three decades since 1990 are reversing. Globalisation, which once acted as a structural dampener on prices, is giving way to friend-shoring, industrial policy, and the construction of parallel economic systems. The US and China are no longer merely trade rivals; they are building distinct technological, financial, and geopolitical ecosystems.

Meanwhile, the demographic dividend that powered the great labour force expansion of the 1980s is exhausting itself. Aging populations in advanced economies are pushing up healthcare costs, shrinking the workforce, and pressuring sovereign balance sheets.  Immigration, which briefly offset these trends, is now being curtailed by a fierce populist backlash across developed markets. The pushback against immigration is destroying supply of workers at precisely the moment AI wealth effects are stimulating demand, making the labour market easier to overheat.

Constructive for now, but humble about tomorrow

None of this means investors should retreat to cash. In the near term, the landscape remains supportive. US underlying growth is tracking at 3.2% for Q2 2026. Profits are rising, hiring is broadening, and AI capex demand continues to outpace supply. Risk assets, for now, have the wind on their backs.

However, the range of plausible long-term outcomes is extraordinarily wide. Will AI prove disinflationary or inflationary? Will productivity gains be broadly shared or concentrated? Will fiscal trajectories stabilise or spiral? Will deglobalisation prove a modest friction or a structural rupture? Even people with historically reliable crystal balls find this a challenging environment.

This is precisely why Fed’s Warsh is right to resist the temptation of forward guidance and high-conviction long-run forecasts. As the Institute of International Finance said: Today’s world is “characterised by wider confidence intervals, more frequent shocks, and greater uncertainty about the underlying structure of the economy”. In such an environment, communicating how the Fed will react to emerging trends and the risks around them is more valuable than pretending to know the destination.

For investors, the lesson is clear. When the distribution of outcomes is this wide – spanning AI-driven abundance and fiscal crisis, productivity renaissance and stagflation risks, a new globalisation and a fragmented world – the rational response is not to bet on one scenario. It is to build all-weather portfolios: diversified across geographies, asset classes, and inflation regimes. These are portfolios built for resilience in both boom and bust, and positioned to capture upside while surviving the tail risks that today’s complacent markets may be dangerously underpricing.

The near term looks fine. The long term is genuinely uncertain. Warsh is right to say so – and investors would be wise to listen.