August 2026 delivered an increasingly unusual macroeconomic picture.

In the United States and Australia, central banks continued to signal that the inflation fight is not finished. At the same time, the technology sector pushed ahead with extraordinary investment in semiconductors, data centres and AI infrastructure.

The result is a market operating at two very different speeds.

The Federal Reserve entered August with the federal funds rate at 3.50%–3.75% following its late-July decision to hold rates.

The minutes from the July FOMC meeting, released in August, reinforced policymakers' continued focus on inflation.

At Jackson Hole, Fed Chair Kevin Warsh maintained that inflation remains too high and emphasised a data-dependent approach to monetary policy. Importantly, however, Warsh did not characterise the U.S. economy as materially weakening. He described the labour market as stable and broadly consistent with full employment, with unemployment at 4.1%, while noting that broader financial conditions were difficult to characterise as restrictive.

That distinction matters.

The Fed is confronting persistent inflation without clear evidence that current financial conditions are sufficiently tight to guarantee its return to target. This leaves policymakers with limited room for complacency—and markets with continued uncertainty over the path of rates.

Australia faces a similar challenge.

The RBA kept the cash rate at 4.35% in August, with inflation remaining above its 2–3% target.

For Australian households, businesses and property investors, the implication is straightforward: investments should work under today's financing conditions rather than depend on aggressive future rate cuts to make the numbers work.

If monetary policymakers remain cautious, the technology sector is showing little hesitation in deploying capital.

Nvidia's August results provided perhaps the clearest evidence.

Its data-centre business generated approximately $89 billion in quarterly revenue, up 117% year over year, illustrating the extraordinary scale of demand for AI compute infrastructure.

The significance extends beyond Nvidia.

Every GPU sold ultimately connects to a much larger investment ecosystem: semiconductor fabrication, networking equipment, data centres, cloud infrastructure, electricity generation and increasingly specialised AI hardware.

Apple's previously announced agreement with Broadcom, unveiled in July, provides another example of this broader semiconductor investment cycle. Apple committed to significantly expanding purchases of U.S.-manufactured Broadcom components as part of its domestic supply-chain strategy.

The AI race is therefore becoming more than a competition between software models.

It is becoming an industrial infrastructure cycle.

This is where August becomes particularly interesting.

Monetary policy remains cautious, yet AI capital expenditure remains highly expansionary.

Those forces can coexist because many of the companies driving the AI buildout possess enormous cash flows, strong balance sheets and considerably less sensitivity to borrowing costs than households or smaller businesses.

But spending alone does not create shareholder value.

Every new data centre, GPU cluster, custom accelerator and power agreement ultimately needs to generate an economic return.

That moves the AI investment debate into its next phase.

The question in 2024 and 2025 was:

How big could AI become?

Increasingly, the question for 2026 and beyond is:

Who captures the return on the hundreds of billions of dollars being invested to build it?

Chip designers capture semiconductor economics. Foundries manufacture the silicon. Networking companies connect the infrastructure. Cloud providers monetise compute. Utilities and infrastructure owners supply the electricity. Software companies attempt to convert all of it into productivity.

Somewhere along that chain, however, customers ultimately have to generate enough economic value from AI to justify what is being spent.

That is the number investors should be watching.

August did not weaken the long-term AI investment thesis.

It did raise the standard required to justify valuations.

We remain constructive on the infrastructure layer of AI, particularly businesses positioned around compute, semiconductor manufacturing, networking and hyper-scale cloud infrastructure.

At the same time, persistent inflation and uncertainty around monetary policy reinforce the importance of valuation discipline.

AI may be a generational technology cycle. That does not mean every AI-related asset is a good investment at every price.

Heading into Q4 2026, we are watching two questions closely: whether inflation gives central banks room to ease—and whether the extraordinary AI infrastructure cycle begins producing equally extraordinary returns.