Two more years, or two more weeks?

The S&P 500 is presently on track for a 20% return for the third year in a row. A so-called three-peat is unusual. But there was almost a five-peat in the late 1990s. That run was spurred on by the first internet revolution. The artificial intelligence (AI) revolution is behind the latest bubble. It’s helping equity markets everywhere and that, along with easing monetary conditions and an absence of fiscal tightening, is keeping the growth plates spinning. For now, the momentum is strong, wealth is being boosted, and tariff and geopolitical risks are fading. The party keeps on going. However, if AI is in any respect a bubble, the hangover is going to be massive.

• Key macro themes – stunning wealth effect; booming credit and no austerity
• Key market themes – intensified US equity concentration; brewing credit concerns

Bullish themes

 There are three themes driving economic growth. The first is the wealth effect resulting from equity markets’ strong performance. The second is the easing of monetary conditions, which is underpinning a credit boom. The third is that fiscal policy is far from restrictive. The wealth effect is supporting consumer spending and the ability of corporates to expand through acquisition and capital spending. Monetary easing is supporting a buoyant credit backdrop, evidenced by heavy issuance and tight spreads in corporate bond markets. Fiscal expansionism is a result of the triumph of populism and the need for security, and is channelling public money into areas like defence, infrastructure and technology. The overall macro-narrative remains bullish – at least the one derived from financial market performance.

Richer 

The wealth effect is stunning. According to Bloomberg data, the market capitalisation of the MSCI World Equity index has increased by $12trn (16%) in 2025. In the US, the S&P 500 index has added around $7.7trn. The top eight stocks in the index (six of the Magnificent 7 plus Tesla and Broadcom who keep swapping seventh place), have accounted for almost half of that. Compared to the estimate of US nominal second quarter (Q2) economic growth, these stocks have a market value equivalent to around 70% of the economy, up from 12% a decade ago. Look at Bitcoin as well. Not as widely held as stocks, of course, but its market capitalisation has increased by around $370bn this year. Investors in gold have also had a sparkling time (the US dollar price of gold is up 55%).

Technology drivers 

AI and blockchain are the technologies driving markets. Hardly a day goes by without US technology firms announcing mutual multi-billion-dollar investment deals, mostly with each other. The circularity of this has been pointed out – AI developers invest in computing power providers who in turn invest in chip manufacturers who then take stakes in AI developers. The AI race is creating a land-grab – for talent, for capital, for data centre infrastructure, for water and electricity and, of course, for customers. The numbers involved are phenomenal and inflationary, as the sector is creating huge demand for what it needs to grow. It’s not surprising the utility sector is the third best performing S&P 500 sector this year (and still trades with a price-earnings valuation well below the Information Technology and Communication Services sectors). 

More money, more spending 

The AI boom effect on equity markets is boosting household wealth. According to the Federal Reserve’s (Fed) Flow of Funds data, net household wealth increased by $5.3trn in the six months to the end of June. As I wrote recently, this tends to benefit higher income cohorts, adding to income inequality, albeit still boosting consumer spending. Pressure on lower income households from the government shutdown, from caution in hiring, and from rising prices for consumer goods (especially those with import content) means there are very different consumer economies existing side by side. Wealth accumulation is happening in a concentrated group of assets and for a concentrated group of capital owners. It is boosting growth but potentially creating social and political risks down the line.

Credit boom 

Despite the absence of official US economic data, the market is still pricing in 125 basis points (bp) of interest rate cuts from the Fed between now and the end of 2026. The implicit assumption here is that tariff-related inflation will be temporary giving the Fed the room to provide monetary stimulus (also encouraged to do so by the US Administration and its nominations to the Board of Governors). Higher asset prices and lower rates are stimulating credit growth. Banks’ lending standards have eased in recent months and loan growth is strong. The outstanding amount of corporate bond debt in the ICE/Bank of America investment grade index has gone up by $325bn this year (reflecting higher bond prices and strong levels of issuance). The same outstanding face value in the high yield market has increased by $64bn and there is no doubt that private credit flows have been extremely strong. The most recent narrative around the bond market has been the rapid increase of fixed income instruments, including asset-backed bonds, to fund investment in AI. None of this suggests an economy in immediate danger of recession. It also adds to the view of a monothematic US economy.

No fiscal austerity 

In government bond markets the vigilantes are restless. There is a general discontent with how public finances are being managed. Populism and the associated political fragmentation tend to work against governments taking hard decisions on taxes and spending – note the current situation in France and the UK. The US variety of populism has been able to shape the fiscal stance in favour of higher income taxpayers and higher deficits. There is a bit of a stalemate in the bond markets with long yields stabilising as investors wait for budget news in the UK and France and being placated in the US by the prospect of lower Fed rates and ongoing growth.  

The Organisation for Economic Co-operation and Development (OECD), in its June 2025 Economic Outlook, warned that governments need to focus on debt sustainability. However, it conceded that government deficits will remain sizeable (OECD aggregate budget deficit of 4.7% of GDP in 2026) and both the US and Germany will experience positive fiscal stimulus. We can possibly add Japan to the list in the wake of the election of Sanae Takaichi as head of the LDP and Japan’s next Prime Minister. Pressure is on governments to raise defence spending and to support investment in renewable energy, digital infrastructure and other critical technology. There is little scope for deep spending cuts. We are far from the austerity years of the late 2010s.

Macro thematic momentum 

These macro themes have momentum. They underpin expensive valuations in equities and credit. Higher valuations support strong sentiment – folks are making money. It is going to take something big to upset markets. 

The question as to whether all the money being spent on developing the infrastructure for AI will generate economy-wide increases in productivity remains unanswered. As does whether the firms at the forefront of the AI arms race that are burning through cash and taking on more debt will fully monetise the investment through persistently high earnings growth. It seems like a bubble, and more and more people are saying it is, but what will burst it, is impossible to say.

For now, the Donald Trump administration appears willing to tolerate the increased wealth and power of the private firms developing AI. But part of AI is going to be state-security critical. Politically, is it going to be acceptable that all that power and wealth and control of the technology is concentrated in the hands of a few private firms? The combined revenue of the top eight stocks is the equivalent of 7.5% of US GDP and growing. The US is increasingly becoming a huge bet on AI. The risks of concentration and disappointment are clear. I also wonder whether the rest of the economy (non-technology enterprises and households) are even going to be able to afford the necessary investment in AI to make the productivity dream a reality. There are other limits to growth – the ability to power and cool data centres being one of them. Is it too dystopian to imagine a world of rationing of electricity usage so that the great AI machine can be fed?

Lending

The credit boom should be a cause for some concern too. I have talked a lot about how tight spreads are. I still hold the view that most corporate bond fundamentals are solid, and excess returns can stay positive. But it is not usually the investment grade markets where credit problems start. It is in more leveraged structures in the high yield market or in leveraged loans. Many are concerned about private credit where leverage is higher, and interest coverage is lower. A more stubborn Fed, in the face of higher US inflation, is a risk as both underlying yields and spreads could move higher if the market is forced to revise its expectations on monetary policy.

FOMO beats Cassandra (most of the time)

I’ve been in the game long enough to know that being a Cassandra generally does not pay off. The fear of missing out (FOMO) optimist has a stronger hand than the perpetual doomster. But occasionally there are bear markets. A lot of good things need to keep happening to sustain the current momentum. Protecting investment portfolios is tempting. Being aware of concentration and the risks is key. As I said, the S&P 500 is on track for a price return of 20% again this year. That would be the third consecutive year. Since the 1940s there has not been a run of three consecutive years of price gains. But there has been a run of four – between 1995-1998 – and nearly a run of five years, as the market increased by 19.5% in 1999. That was followed by three years of negative returns. It was the first internet boom. This is the second. It could go on for another two years or it could end in wealth destruction and recession. The long-term average price gain of the S&P 500 since 1950 has been 8.12%. One might want to hedge against mean reversion.