When the Money Goes Nowhere — and Everywhere at Once
Today, I heard the phrase used by Sunaina Sinha, who described the current state of IPOs and the flood of capital into equity markets as the "Party of Capital". This resonated deeply with me. I may sound like a broken record to some colleagues, as I keep returning to two numbers that should not coexist in the same economy.
The US M2 money supply sits at a record $22.8 trillion. The velocity of that money — the rate at which a dollar actually circulates through the productive economy — is 1.41. That is near a 65-year low. For context, from 2002 to 2008, velocity ranged from 1.97 to 2. The money is flooding the market assets, simply not working for consumer spending. And to prove that, three companies are entering public markets at a combined valuation exceeding $4 trillion. SpaceX has filed its prospectus targeting $1.75 to $2.3 trillion. OpenAI is targeting $850 billion to $1.1 trillion. Anthropic, following a $30 billion Series G led by Singapore’s GIC and Coatue, is eyeing up to $900 billion.
Here is a statistic that should give most investors pause – the current cohort of AI and space tech IPOs represents the most highly valued pipeline of concentrated enterprise wealth in financial history. At a projected $2.2 trillion, this handful of companies is worth more than the total capital raised by every dot-com IPO from 1995 through 2000 combined ($245 billion) and rivals the aggregate market capitalisation of the entire five-year tech boom at its absolute peak. The comparison is not rhetorical. It is structural.
To add to the challenge, none of these three companies is currently profitable. SpaceX’s 2025 consolidated accounts show revenue of $18.7 billion against CapEx exceeding $20 billion, with losses approaching $5 billion. It is clear that the market is not pricing in earnings. It is pricing an option value on a future that has become a vision purely on the back of decades of cheap money.
The post-COVID spate of indiscriminate and uncontrolled fiscal spending contributed to an unprecedented 40% expansion of the money supply over the two years between 2020 and 2022. The Fed’s balance sheet went from $4 trillion to $8.9 trillion. Three years of quantitative tightening, which formally ended in December 2025, managed to remove ~only $2.31 trillion, or 25.7%, of peak assets. The Fed’s balance sheet still sits at around $6.6 trillion. However, during the same period, CARES, ARP, IRA, IJA, and the CHIPS Act alone injected $7 trillion into M2, sustained by deficit monetisation, with funds issued as subsidies, tax credits, and grants that, due to regulatory and human capital shortages, sit idle in Treasury accounts, remaining locked as static M2. A high Fed funds rate environment incentivised corporate and retail capital to park in high-yield MMFs rather than circulate in the real economy.
The QT experiment was clearly insufficient to address the structural excess, whilst the federal and state governments continued unconstrained deficit spending. The central bank created reserves that permanently changed bank behaviour. Those reserves funded runnable liabilities — demand deposits, credit lines — that persisted even as the Fed withdrew them. The result is a liquidity-dependent financial system that now resists normalisation.
Convenience yields across the US, UK, Germany, and Japan have all turned negative by the end of 2025, signalling an eroded trust in the risk-free status of sovereign bonds under QT conditions. This transition to negative convenience yields marks the end of the "sovereign subsidy" era. For decades, G4 governments enjoyed a financial discount because the global financial system desperately needed their paper as plumbing. But by oversupplying the market while simultaneously ending central bank backstops via QT, sovereign bonds have been demoted from "pristine collateral" to "duration risk that must be digested." The global market still trusts the underlying currencies, but it now demands extra compensation to hold governments' uninhibited issuance.
What makes this M2 expansion even more damaging is that it didn’t contribute to wage increases. It didn’t build infrastructure. It accumulated in asset markets, where returns were visible, and velocity became irrelevant. I feel that policymakers now treat QE as a “crisis-management tool”, assuming the exit will be a manageable operation. I think this is a false and dangerous assumption. The entry was fast, clean, and politically easy. The exit will be slow, asymmetric, and politically toxic.
Add to this the unprecedentedly large anticipated IPOs of 2026; I don’t believe they are the result of organic economic expansion. I believe they are more a product of two decades of compressing discount rates. Private market investors — sovereign wealth funds, venture capital, private credit funds, and late-stage growth equity — have competed to deploy capital into a narrow set of high-conviction AI and space technology platforms. Sovereign wealth funds globally now manage a record $15 trillion in assets, having deployed almost $66 billion into AI and digitalisation in 2025 alone!
GIC’s $30B round in Anthropic, Mubadala’s $32.7 billion across 40 transactions last year and PIF’s $36.2 billion in 2025 – this is patient capital backed by sovereign balance sheets. I feel the current AI, Space IPO supercycle is ‘partly’ a GCC and Singapore sovereign capital placement — hydrocarbon and export surpluses being repositioned as compute infrastructure, with the US public market being the platform.
The FDI picture, too, is consistent with this narrative. The US is simultaneously the largest source and the largest destination of AI capital. That closed-loop structure is not a diversified system. It is a concentration. US greenfield FDI doubled to $237 billion in just six months — driven almost entirely by AI-related investments in semiconductors and data centres. Inbound FDI into the US was $89 billion in Q3 2025, the highest since Q2 2023. However, globally, overall FDI fell 3% in 2025. Industry and infrastructure projects fell 15%. Greenfield investments excluding tech declined 17%. SDG-related investments fell 10%. Right now, capital is excessively focused on AI infrastructure in the US, whilst withdrawing from productive economic capacity elsewhere. On the other hand, US companies accounted for 45% of total global AI-related FDI in 2024. Inbound AI FDI was $32.5 billion in 2024, approximately 30% of global AI investment flows.
Contrast this with the fact that 58% of Americans carry credit card balances they cannot pay off. US credit card debt hit a record $1.28 trillion in Q4 2025. In March 2026, revolving debt surged 9.1% as consumers absorbed the oil shock from the Iran conflict — fuel costs up 28.4% year-on-year, fuel oil up 54.3%. Headline CPI reached 3.8% in April 2026, the highest since May 2023. Core CPI edged up to 2.8%. Consumer credit outstanding jumped 5.8% annualised in March, but after adjusting for inflation, real consumer credit contracted 0.4%. The consumer is sustaining spending through debt, not income. Personal consumption expenditures have held up. Disposable personal income has flatlined. I have seen this enough in MENA and other emerging markets to know what this pattern looks like at ground level: an excellent top-line growth until the month it doesn’t. Delinquency on credit cards was already at 7.18% in early 2025, well above the pre-pandemic 6.36%.
The equity market, in my view, is pricing the future, whilst the consumers are borrowing against the present. They are two different economies, running in parallel, connected only by the illusion of aggregate GDP. What worries me is that this concentration is structural. Pension funds are now deeply embedded in this architecture. S&P Global found that US state pension funds nearly doubled their private investment allocations over the past decade. 18% of pension funds surveyed by Ortec Finance say they lack sufficient liquidity to withstand adverse scenarios. 80% say unfunded commitment risk poses a significant challenge to defined benefit schemes over the next three years.
This leaves the retail investor doubly exposed. First, through pension fund participation in the market boom via private credit, and secondly, through equity indices that will include OpenAI and SpaceX valuations, factored into 401(k) allocations without any active choices available. The dot-com parallel is structurally closer because of concentration risk. The five largest companies now hold approximately 30% of the S&P 500. In March 2000, the Nasdaq tech sector held ~35% of the S&P. The exposure for passive investors, if sentiment shifts, is comparable.
What would worsen the outlook, if core CPI at 2.8% continues to drift higher — driven by energy costs embedded in the Iran conflict supply shock — the Fed will face a rate path that punishes a consumer already at 7%+ delinquency. QT at scale, per BIS modelling, reduces GDP by approximately 0.65% and cuts inflation by 45 basis points per meaningful tightening cycle. The resulting pain will be felt across the economy.
Lagarde, too, faces an additional layer: eurozone consumers are under the same oil shock, with German and French household sentiment deteriorating, but the ECB’s mandate is narrower and its political economy more complex. The convenience yield on German Bunds turned negative by the end of 2025. The ECB began balance-sheet normalisation in March 2023 and is now watching those yields erode further as bond supply increases.
We are left with two central banks, each with chronically large balance sheets, facing rising inflation and slowing GDP growth, and both facing strong political constraints that limit aggressive tightening to address the M2 excess. The reality is QT of sufficient scale to meaningfully reduce M2 has never been executed by any major central bank without triggering a financial stability event. Japan spent 30 years trying. The Fed’s three-year experiment trimmed 25.7% and then stopped. The excess supply of post-QE money is not a problem that monetary policy alone will solve.
The velocity problem requires fiscal intervention, not monetary tinkering. A money supply of $22.8 trillion, with velocity at 1.41, means the productive economy is operating on approximately $32 trillion in nominal transactions. If velocity returned to its pre-GFC level of 1.97, the same money supply would generate the equivalent of $45 trillion in transactions — without printing a single additional dollar. The policy levers that would actually move velocity are directed credit — tax incentives, public guarantees, or direct lending for SME investment, housing construction, and green infrastructure; incentivising reinvestment into productive assets rather than further asset appreciation; public investment mandates attached to SWF and pension fund allocations as a condition of regulatory approval; and structural reform of the private credit market, requiring mark-to-market valuation, public disclosure of underlying assets, and liquidity buffers proportional to pension fund exposure.
But none of these are simple. All of them require more political capital than what most OECD governments currently possess.
To summarise, I don’t think we are replicating 2008 or 2000 precisely. The instruments are different, the balance sheets are bigger, and the companies being valued are — in some cases — generating real revenue. But the architecture underlying this moment has the same load-bearing flaw as both prior crises: leverage and concentration in a narrow set of assets, with institutional intermediaries — pension funds, private credit managers, index-passive allocators — creating systemic exposure for retail investors with no visibility into it.
The SpaceX IPO, if it lists at $1.75 trillion in June 2026, will be immediately absorbed into major indices. Every passive investor holding a US equity ETF will own it within weeks, at the price the market has been asked to validate — not at the price that reflects a company with $18.7 billion in revenue and losses approaching $5 billion.
What I keep coming back to is a question that sits uncomfortably with both the bulls and the bears. Not whether the technology is real — it is. Not whether the valuations will hold — some will, some won’t. But whether the mechanism by which we have funded this wave of innovation is one we would have designed deliberately, with full knowledge of the distributional consequences. Trillions in central bank balance sheet expansion. Record consumer debt. Velocity at a 65-year low. Pension funds overexposed to opaque private assets. And three rockets heading for public markets with a combined price tag larger than Germany’s GDP.
The money went somewhere. The question worth a global dialogue is about who exactly is holding the bill.
More in this area
Midterms and AI expenditure
A few people have asked me whether the midterms put the brakes on the AI buildout. I went back through the data, and my view is that 'Bluenami' :-) will change the pace and the cost of the AI buildout. They will not change its…
Debt, Inflation, and the Narrowing Room for Policy—The Rising Cost of Running the State
I listened to Andy Burnham’s first Labour conference speech as Prime Minister, as he set out a larger role for the British state: a national care service, more council housing, changes to pensions and greater public involvement…
The Rupee’s Catch‑22: How India’s Weak Currency Became Its Own Worst Enemy
I am watching the current debate raging on social media on the decline of the Indian currency with some intrigue. From 1994 to 2024, I have seen the Indian rupee slip from roughly 31.37 to the dollar to about 83.51, a…