June 24, 2026 — 11:59am
Blame Kevin Warsh for a plunge in technology related shares over the past few trading days that wiped out nearly $US1 trillion ($1.45 trillion) from the value of Elon Musk’s SpaceX.
It is no coincidence that the selloff of tech shares – particularly semiconductor makers, but more generally everything artificial intelligence-related – started on Thursday.
That was the day the US Federal Reserve Board’s Open Market Committee’s first meeting chaired by Warsh ended.
The rate-setting committee provided a clear message – buttressed by a “hawkish” tone from Warsh himself – that it was more likely that the Fed would be raising US interest rates by the end of the year than lowering them.
In fact, everything that flowed from that meeting and the announcement of Warsh’s five taskforces to review the way the Fed communicates and operates pointed to a more conservative, smaller, less interventionist central bank in future.
There was something coincidental about the timing of that Fed meeting, with former long-serving Fed chairman, Alan Greenspan, dying this week.
Warsh’s plan for a smaller Fed balance sheet that holds fewer bank reserves would pass much of the responsibility and risk of supplying liquidity to markets.
Instead of the “Greenspan put” – a conviction borne out of the way the Greenspan Fed intervened to put a floor under financial markets during the Asian financial crisis, the Long-Term Capital Management collapse, the bursting of the dot-com bubble and the 1987 sharemarket crash – markets would have to price risk into asset prices.
Greenspan’s tenure as chair was characterised by a willingness to bail out financial markets, loosen monetary policy and financial deregulation. If Warsh is true to his stated convictions, his Fed will be very different (and a major disappointment for Donald Trump).
Technology stocks, especially AI-related stocks – and most particularly SpaceX with its stratospheric valuation that capitalises a vision rather than prospective revenues or earnings – are vulnerable to even the hint of a rate rise and even more vulnerable if investors believe the safety nets they’ve relied on for decades may soon be yanked.
That’s partly a function of the market’s maths and partly due to the convoluted and vulnerable way the AI revolution is being financed. The maths is conventional; the financing definitely isn’t.
Conventionally, investors value companies by discounting their forecast future cash flows by a risk-free rate, which is generally the yield on 10-year government bonds. The higher that yield, the lower the net present value of those future cash flows.
The US 10-year bond yield has been creeping up ever since Donald Trump regained the presidency last year but, until last week, there was no expectation that the Fed itself would target higher rates.
After the Fed meeting, yields rose again by a few basis points and, more pertinently, futures markets priced in at least one 25 basis point rate rise before the end of the year.
Bank of America economists who, before the meeting, had been forecasting no rate movements this year, are now saying there could be three.
That might be extreme, but it illustrates how abruptly the mood and expectations have shifted.
For the big tech companies, trading on abnormally high multiples of their revenues and earnings – in SpaceX’s case, when it was trading at 160 times last year’s revenues – even a modest shift in interest rate expectation has a magnified impact on their valuations.
The worst hit sector of the market was semiconductor stocks, with the Philadelphia Semiconductor Index plunging almost 8 per cent this week. The tech-heavy Nasdaq index is down 3.5 per cent since the Fed meeting and the broader S&P 500 just under 2 per cent.
The “Magnificent Seven” stocks – Alphabet (Google’s parent), Amazon, Meta (Facebook), Microsoft, Nvidia and Tesla (now displaced in the top seven by SpaceX) – that account for more than a third of the US market’s capitalisation and have driven the market this decade, with an acceleration as the AI story gained traction and momentum, are down nearly 4 per cent on average in the past few days.
SpaceX, which debuted on the market last week with a market capitalisation of $US2 trillion, a valuation that peaked at $US2.99 trillion, is now valued at $US2 trillion again.
While market maths and a post-SpaceX broader reappraisal of how much hype was being built into its and other AI companies’ valuations might explain why tech shares have reacted to the Fed’s evolving outlook for rates, the share prices of the big tech companies had started retreating from record levels very late last month.
That seems to have been driven by a recognition that the forces driving the market – the massive investments occurring in AI models and the infrastructure for training them and a demand for semiconductors that was outstripping supply – were stressing a different equation.
Any crack in the confidence investors have had in all things AI-related and the notion that AI valuations can only ever increase, however, represents a threat to the sector and the broader market.
The investment in AI, estimated at $US750 billion by the handful of hyperscalers this year and more than $US1 trillion next year and beyond, is accelerating and growing at a faster rate than the AI-related revenues the investments are generating, even though the revenues of the sector’s leading protagonists are growing at exponential rates.
Such is the demand for capital that SpaceX listed and raised $US85.7 billion and immediately hit the debt market for $US25 billion.
In recent days, Amazon has issued $US80 billion of new debt and Nvidia $US25 billion, while Alphabet raised $US80 billion of equity. These are companies with massive cashflows from their traditional operations, companies more used to announcing share buybacks than bond issues.
The two leading pure AI start-ups, Anthropic and OpenAI, have also filed prospectuses for their own hoped-for trillion-dollar-plus sharemarket listings and capital raisings, although they’ll now be monitoring the current market anxiously, worried that they might have missed the boat.
The divergence between the growth in AI spending and revenue growth and the losses from AI investments are unsustainable.
Either the already impressive revenue growth being posted by companies like Anthropic (an annual revenue run-rate of $US14 billion in February exploded top $USb47 billion last month) has to grow even faster, or the growth rate in investing will need to be cut back.
Until now, equity – much of it private – and, increasingly, debt have filled in the widening gap between revenue and spending.
If the markets have become nervous about the sustainability of AI finances, that capital will become either more expensive, less available or probably both.
SpaceX’s debt raising was notable because the overwhelming demand was for the short-duration, less risky, debt. Yields on the debt were higher than for companies with similar ratings. Investors do seem to be a little more conscious of AI risks than they were earlier in the year, when there was uncritical and open-ended investor financial support for all things AI-related.
The AI sectors’ funding structure is incestuous, with lots of cross-shareholdings between the key players and back-to-back contracts for equity, chips and data centre capacity. It’s been described as “circular finance.”
If the AI companies were to lose access to equity markets, there is a risk that the entire sector would implode. The heavyweight techs with vast non-AI cashflows, would survive with a lot of expensive scarring, but slabs of the sector would disappear.
That’s probably not going to happen, at least not yet, and the wider sharemarket might be experiencing just a modest correction after a very big run up. Any crack in the confidence investors have had in all things AI-related and the notion that AI valuations can only ever increase, however, represents a threat to the sector and the broader market.
Warsh and the “reformed” Fed that he envisages, could provide that threat.
The Market Recap newsletter is a wrap of the day’s trading. Get it each weekday afternoon.

















