Monday Digest
Posted 28 September 2026
Waiting for Goldilocks
Bond yields keep rising. The US 10-year yield is at 5.2%, its highest level in nearly two decades. It got there last week, taking other nations’ bonds with it.
Although energy prices continue to be a negative, it is growth expectations, rather than oil prices, that are moving yields the most. US business sentiment surveys were remarkably strong. That’s awkward for the Federal Reserve: the US economy is near capacity, so extra demand will push up prices. Markets now expect another rate rise next month.
The Fed’s hawkish turn pushed short-term rate expectations up faster than long-term yields, flattening the yield curve and tightening financial conditions. Stocks suffered, with the now-typical exception of AI-driven tech firms. The Fed’s problem is that the economy’s AI motor seems undisturbed by higher rates, despite the rest of the economy struggling. Bond investors yearn for the slow-but-steady ‘goldilocks’ growth that characterised the 2010s.
A friendly Trump-Xi summit produced little of substance but preparatory talks extended a tariff truce until January – conveniently after the US midterms. US-China détente removes potential disruption for US consumers, though it does little to address Trump’s main electoral headache: fuel prices.
With big tech’s capital demand crowding out other investment, the AI trade is single-handedly carrying markets. AI firms have amassed vast cash piles to build datacentres but, curiously, datacentres seemingly aren’t being built and computing capacity is stalling. Even stranger, constrained computing capacity isn’t resulting in higher compute prices.
That suggests current computing capacity is ample. And since AI firms have already raised unspent billions, you’d expect them to borrow less going forward. That’s one path back to a goldilocks environment – AI spending cooling along with the rest of the economy, allowing central bankers to back off. The other route is energy price reprieve, but we won’t bank on that. Without either of those, market breadth will probably keep narrowing. That doesn’t preclude investment returns, but it makes them more fragile.
‘Pump and dump’ turned Ponzi
Türkiye’s stock market had a storming start to the year but has since hit serious turbulence. High-profile funds like Tera Portföy and Pusula Portföy have blocked investor redemptions after running into liquidity trouble tied to a handful of thinly-traded stocks. That caused a broader sell-off and whispers of financial collapse.
The funds in question had mystifyingly good returns in recent years, seemingly thanks to market manipulation. Türkiye’s free float is unusually small – just 33.5% on average – and fund managers exploited that by trading illiquid shares between themselves to push up prices.
Tera Portföy is the poster child. It invested heavily into a handful of closely related stocks and used leverage to amplify its bets. One of those was its own parent company, creating a feedback loop where the fund’s gains lifted the parent’s share price, in turn boosting returns. At one point, Tera stock made up 99% of the fund’s holdings.
It worked for a while; Tera’s flagship fund was 660 times its 2022 listing price at one point. But the same illiquidity that inflated the fund also sank it once outflows overtook inflows. The government has since ordered an $18bn liquidation and arrested Tera’s chairman.
Many long suspected the fund’s returns were too good to be true – but as long as they kept coming, investors kept investing.
The trigger for the collapse was a tightening of rules on illiquid holdings, and the catalyst for that seems to have been MSCI threatening to demote Türkiye from ’emerging’ to ‘frontier’ market status over transparency concerns.
We see little sign of contagion beyond Türkiye, and no obvious banking crisis brewing. But the affair is a reminder of why mature, well-regulated markets like the G7’s matter: strong oversight doesn’t eliminate the temptation to chase self-reinforcing gains, but it does make Ponzi-like structures much harder to sustain.
SoftBank’s big bond bet
SoftBank has launched $11bn of junk-rated bonds, reportedly to fund its OpenAI investment ahead of the ChatGPT maker’s stock market listing. The Japanese investment group will pay a 9.75% yield on the debt’s longest tranche, underlining its confidence in the company. SoftBank has already committed $65bn for a 13% stake in OpenAI.
SoftBank and its outspoken founder Masayoshi Son have a history on big tech bets – the most famous being its $20mn Alibaba investment, which turned into $60bn when the tech firm went public. Masayoshi Son has been likened to Warren Buffet, though Cathy Wood’s ARK Innovation ETF is a better comparison, given Son’s growth-over-value approach. SoftBank is similar to Berkshire Hathaway in the way it leverages investment, however, shown by its willingness to pay a near-10% bond coupon.
SoftBank’s leverage mirrors the rising leverage in the tech sector itself. Once-rich AI firms have become incredibly capital-intensive. Oracle, whose credit default swaps hit record highs earlier this year, exemplify this. That debt has helped fuel earnings growth and share prices. SoftBank’s bond shows the leverage increase is happening at both the company and investor levels – and demand for the bond shows the process isn’t slowing down.
Tech firms still have low leverage ratios relative to everyone else. Indeed, that’s why they’re able to borrow while other firms and households aren’t. The effect on overall borrowing costs means AI companies are crowding out other investment in the way governments are sometimes accused of.
Things aren’t all fine for AI companies themselves: higher leverage makes them more interest rate-sensitive just as rates are climbing, and it makes earnings distortions (like Amazon’s earnings gain from its SpaceX holdings) more likely. Big tech’s credit spreads have been more volatility lately too.
SoftBank’s big bond isn’t an alarm bell, but investors should keep an eye on AI leverage.