NextFin News - DBS Group Chief Investment Officer Hou Wey Fook says AI-led earnings growth continues to underpin major technology stocks, while investment-grade bonds yielding above 5% offer an attractive source of income and portfolio stability. The remarks, made in an interview aired October 5, 2026, pair two positions that look complementary on the surface but sit on a single, fragile assumption: that the policy framework anchoring those bond yields remains intact.
The tension is deliberate, and it is the story. Hou's barbell — growth on one side, income on the other — only works if the income leg is genuinely stable. In his own earlier commentary this year, he identified the single biggest threat to that stability: political pressure on the Federal Reserve. The question his latest positioning raises is whether the very asset he is recommending as ballast is exposed to the risk he has already warned about.
The Barbell: AI Earnings and 5% Income
On the equity side, the DBS CIO's argument is a direct rebuttal of the "AI bubble" crowd. The optimism around artificial intelligence is not built on exuberant expectation detached from fundamentals, but on earnings growth that is already materialising and is funded differently from the dotcom era. During DBS's first-half 2026 outlook briefing in January, Hou drew the contrast sharply: "The crash during the dotcom bubble was entirely due to exuberant expectation of earnings growth, which did not materialise," he said, pointing to an 88% collapse in earnings growth during that period. "The story is very different today."
The difference, in his telling, is the funding source. The capital expenditure of large technology companies today — Microsoft, Google, Meta and Amazon chief among them — is financed out of operating cash flows rather than from the companies incurring massive debt. DBS notes projections that hyperscalers are set to pour US$1.4 trillion into AI infrastructure between 2025 and 2027. Because the spending is self-funded, the downside is bounded: if the return on that capital disappoints, the damage lands on shareholders through slower growth, not on creditors through a debt spiral. That is a materially different failure mode from 2000.
On the bond side, the pitch is simpler: investment-grade credit yielding above 5% pays investors to wait, and it does so from issuers with strong balance sheets. DBS's own wealth-management commentary this year described the income side of its barbell as investment-grade bonds plus dividend-generating equities, with stable yields in the 4%-6% range in local currency terms and higher in US dollar equivalents. "We do not go into high yield," Hou said in an April interview, preferring to anchor portfolios with predictable income streams. The message is that fixed income has reclaimed the role it lost in the zero-rate decade: a genuine source of return, not just a volatility dampener.
Put together, the two legs form a coherent portfolio story for a high-rate environment. Equities capture the structural earnings growth from AI; bonds capture the income that higher rates made available again. It is the kind of balanced positioning that reads well in a client presentation. The problem is that both legs depend on one variable the market has grown comfortable ignoring.
The Backdrop: A Fed That Is Still Tightening
The first thing to get right is where the Fed actually stands. On September 16, the Federal Open Market Committee raised the target range for the federal funds rate by a quarter percentage point to 3.75%-4.00%, a unanimous 12-0 vote and the central bank's first rate increase since July 2023. Chair Kevin Warsh, who took office in May, framed the move as unavoidable: "The plain fact is that inflation is too high and has been for too long," he said, with the consumer-price measure running at 3.4% year-over-year in August — well above the Fed's 2% target.
The market is not betting on relief. As of early October, Fed funds futures implied a 77% probability that the October 28 meeting ends with rates held at 3.75%-4.00%, and a 23% probability of another quarter-point hike to 4.00%-4.25%. By the December meeting, pricing centres on one more increase, with roughly two-thirds of the probability mass at 4.00%-4.25%. In plain terms: the market expects the tightening cycle to continue, not to reverse. The 10-year Treasury yield sits around 5.28%, and investment-grade credit spreads are near their tightest levels in more than two years — the ICE BofA US Corporate Index option-adjusted spread closed at 79 basis points on September 24, down from levels last seen in late 2021.
This matters because Hou is recommending bonds yielding above 5% as "income and stability" into a cycle that is still hiking. Duration risk — the mark-to-market loss that comes when yields rise — has not yet run its course. A 5% coupon is attractive; it is not a shield against capital losses if the 10-year yield pushes toward 5.75% or higher. The stability Hou is selling is real income, but it is income with a live interest-rate exposure attached. That is the first layer of tension, and it exists even before politics enters the picture.
The Risk the Market Is Not Pricing: A Politicised Fed
Hou has been unusually specific about what he sees as the underpriced risk. In January, he argued that the US is entering an era in which the Federal Reserve may lose its independence, rolling back the 1951 Treasury-Fed Accord that has kept fiscal and monetary policy separate for more than seven decades.
"(US President Donald Trump) is attempting to roll back this accord by pressuring the Federal Reserve through installing a new Fed chairman, as well as FOMC (Federal Open Market Committee) members who are aligned with him in slashing interest rates," he said. Diluting the Fed's independence will "certainly increase" the risk of escalating inflation going forward.
This is not a generic "watch inflation" warning. It is a claim about the regime that sets the risk-free anchor for every asset in the barbell. The risk is not that the Fed cuts too soon — it has just done the opposite. The risk is that it is pressured to stop tightening before inflation is beaten, or to cut prematurely once the political cost of higher rates becomes too high. If the Fed abandons the fight with inflation still above target, the consequence is not a one-off bond rally; it is a repricing of the term premium across the curve. Investment-grade bonds yielding 5% today are attractive only if that yield is not eroded by higher inflation tomorrow. A central bank that loses its inflation-fighting credibility hands bondholders negative real returns even at nominally generous coupons.
The timing adds a second layer of discomfort. Hou's warning was made in January, before the September hike. Since then, the FOMC has delivered a unanimous 12-0 increase despite the political pressure he described. That is evidence for the other side: the institution bent but did not break. The question now is whether the unanimity holds when the next hike is proposed, or whether the pressure fractures the committee as it moves deeper into restrictive territory.
There is an uncomfortable asymmetry in the barbell itself. If the Fed stays independent and keeps tightening until inflation is contained, the bond leg eventually wins: yields peak, capital gains follow, and 5% coupons compound. The equity leg, meanwhile, has to prove that AI earnings can grow through a restrictive policy stance — which is exactly what the dotcom era failed to do. If the Fed is pressured into premature easing, the equity leg may hold up — cheaper money helps growth stocks — while the bond leg, the "stability" side of the portfolio, becomes the source of volatility. The asset recommended as ballast becomes the risk. That is the second-order consequence the barbell pitch does not advertise.
Cyclical or Structural: Separating the Two Legs
The cleanest way to think about Hou's positioning is to split it into its cyclical and structural components, because they point in different directions and carry different holding periods.
The AI earnings story is structural. Three pieces of evidence support that call. First, the funding mechanism is different: dotcom-era capex was debt- and equity-fuelled speculation; today's AI capex is funded from operating cash flow at companies with fortress balance sheets. Second, the demand driver is not a single consumer fad but a broad-based productivity shift across enterprises, cloud providers and sovereigns. Third, the earnings are already visible — Hou expects big tech to deliver "impressive double-digit earnings growth" in the coming years, and the US$1.4 trillion infrastructure build-out is a multi-year commitment, not a one-quarter impulse. A structural shift does not revert on its own; it reverts only if the underlying driver breaks. Here, the break condition is debt-funded capex colliding with missed earnings.
The bond-income story is partly structural and partly cyclical. Structurally, fixed income's income role has been restored: yields above 5% mark a genuine regime change from the 2010-2021 era of near-zero rates, and that is unlikely to fully reverse unless the world returns to secular stagnation and zero-bound policy. Cyclical, however, is the spread compression. Credit spreads at 79 basis points are pricing an unusually benign default environment. DBS itself acknowledged this tension in its second-quarter outlook, noting it had moderately reduced its overweight in bonds and moved the overall allocation to neutral amid tight spreads. In other words, the house view is already trimming the very exposure being recommended to clients.
The duration decision sharpens the point. DBS reiterated a preference for high-quality bonds with shorter duration — two to three years — while maintaining a portfolio duration of five to seven years. That is a barbell within the barbell: lock in today's yields on the short end, keep some sensitivity to a potential rally on the long end. It is a sensible hedge against the scenario where the Fed stops tightening and the curve steepens. But it is also an admission that the long end is not safe.
So the verdict: the AI leg is a structural growth allocation with cyclical valuation risk; the bond leg is a structural income allocation with cyclical spread risk and a structural regime risk sitting behind it. Conflating the two into a single "balanced portfolio" narrative hides where the real exposures are.
The Counter-Thesis: Why the Barbell Could Be Right
The strongest case against this reading is that the evidence already points the other way. Hou's politicisation warning was issued in January. Nine months later, the FOMC he described as under siege delivered a unanimous 12-0 rate hike — the first increase in more than three years — with Chair Warsh publicly framing inflation as "too high." If the institution were truly being captured, the pressure would show up in dissent, in delayed action, or in softer language. It has shown up in none of the three. The 1951 Accord is a convention, not a law, and conventions have inertia; governors and regional Fed presidents have fixed terms, and the committee has shown it can resist political pressure when inflation is the alternative.
On that view, the bond market is right to accept the tightening path, and Hou's warning is a tail risk, not a base case. The 77% probability of holding at the October meeting, with a further hike priced for year-end, reflects data — inflation at 3.4% and a resilient labour market — not capitulation to politics. If the Fed delivers two more quarter-point hikes by December, taking the range to 4.25%-4.50%, the politicisation thesis is wrong and the bond leg is safer than the warning implies. On that scenario, the barbell works: AI earnings compound, coupons compound, and the political noise fades.
That counter-thesis is credible, and it is the base case for most fixed-income desks today. But it depends on two things holding: inflation actually cooling under pressure, and the committee's unanimity surviving the next vote. If core inflation prints at 0.3% or higher month-over-month for two consecutive months while political pressure mounts, the Fed faces the exact trade-off Hou is warning about: keep tightening and risk a political confrontation, or pause and risk re-igniting prices. That is the fork where the barbell's two legs diverge — and where the "stable income" story gets its first real test.
What to Watch
The forward view splits by horizon, and the signals are different at each one.
In the short term — through the October 28 FOMC meeting and into year-end — watch the rate path and the committee's unity. The market prices a hold next month and one more hike by December. If the Fed delivers the hold but the statement or the press conference signals fewer hikes ahead than priced, expect the long end of the curve to sell off and spreads to widen. That would be the first stress test of the bond leg. Conversely, a unanimous decision to hike again would strengthen the independence case and support the stability thesis.
In the medium term — the next two to four quarters — watch AI earnings against capex. The thesis that this is not dotcom rests on cash-flow-funded spending meeting double-digit earnings growth. If hyperscalers report two consecutive quarters of debt-funded capex expansion with earnings missing double-digit growth, the structural call on AI equity breaks down. Conversely, if earnings hold and capex stays self-funded, the equity leg has room to run even at current valuations.
In the long term — beyond 2027 — watch the Fed's institutional posture. The falsifying signal for the risk warning is specific: if the FOMC delivers two more quarter-point hikes by December, taking the range to 4.25%-4.50%, despite public political pressure, the politicisation thesis is wrong and the bond leg is safer than Hou's warning implies. If instead the Committee pauses or cuts while core inflation runs above 0.3% month-over-month, the regime-risk thesis is confirmed, and 5% coupons will not protect bondholders in real terms.
There is also a spread signal worth naming. The ICE BofA US Corporate Index option-adjusted spread sits at 79 basis points. If it widens beyond roughly 120 basis points while investment-grade absolute yields fail to hold above 5%, the "income as stability" thesis breaks — investors would be getting neither compensation for risk nor protection from rate moves. That combination would mark the end of the benign credit cycle the barbell depends on.
The Bottom Line
Hou Wey Fook's barbell is a well-constructed answer to the question of how to invest in a high-rate, AI-driven market. The equity leg is structurally sound; the income leg is genuinely attractive by the standards of the past fifteen years. But the two legs are not independent. They share a single anchor — an independent central bank — and that anchor is the one thing the CIO himself has flagged as under threat.
The investor takeaway is not to abandon the barbell, but to see it clearly: the bond side is not the safe side unless the Fed stays independent. Income above 5% is real money today; whether it stays real money tomorrow depends less on credit fundamentals and more on a political fight over who controls the cost of borrowing.
The takeaway: DBS's barbell works if the Fed stays independent. The moment it doesn't, the asset sold as stability becomes the risk — and the CIO's own warning is the only hedge he has offered.
更多独家洞察尽在 nextfin.ai.
