Research and market developments from 25 September to 2 October 2026.
At a Glance
- The global rates shock is becoming a debt-sustainability question as well as a monetary-policy question. PGIM sees resilient growth, expensive energy and further central-bank tightening putting pressure on government finances, particularly in Europe, while BlackRock notes that long-dated US Treasury yields have reached multi-decade highs. The lower-than-expected US Personal Consumption Expenditures (PCE) inflation readings released on 30 September reduced the immediate pressure for another Federal Reserve increase, but the release also incorporated annual revisions to earlier data and did not remove the structural forces keeping longer-term borrowing costs elevated. [1, 2, 5]
- Asian fixed income has absorbed substantially less of the global repricing. Eastspring Investments found that Asian government yields generally rose much less than US Treasury yields over the previous month, while Asian credit remained ahead of US corporate bonds year-to-date. The correction has improved starting yields and carry, but Eastspring still favours selective exposure rather than a broad duration call. [3]
- AI financing is spreading into public infrastructure and local credit markets. BlackRock estimates that AI-related municipal-bond issuance could reach USD11 billion in 2026, still less than 2% of expected overall US municipal issuance. The important credit question is increasingly who bears the cost of power, transmission, water and other infrastructure required by data centres. [4]
- AI’s second-order beneficiaries are becoming more important. Citi Wealth sees cybersecurity as one such exposure, citing Gartner forecasts that security spending specifically for AI systems could grow at roughly 65% annually from 2026 to 2028, around five times the expected growth rate of overall cybersecurity spending. The attraction is participation in AI adoption without requiring a view on which model or platform ultimately wins. [6]
- Equity exposure is broadening, but quality remains central. In its 29 September bulletin, Citi reported that its Global Investment Council had increased exposure to Japan and US large caps while reducing securitised fixed income at its 24 September meeting. Japan combines improving economic momentum with an unusually strong earnings-revision cycle, while US large caps continue to offer high interest coverage and strong earnings growth. BlackRock separately sees China moving higher up the manufacturing value chain, but stresses that industrial scale does not automatically translate into attractive shareholder returns. [2, 6]
The institutional research reviewed by Hubbis this week centres on a market that continues to generate growth while demanding substantially more capital to finance it. Government borrowing, artificial intelligence infrastructure, energy investment and geopolitical resilience are all competing for funding at the same time. The result is a higher hurdle rate across both bonds and equities.
Yet the implications are not uniformly defensive. Asian fixed income has proved comparatively resilient, corporate earnings remain supportive in several major markets, and the AI capital cycle is creating investment opportunities well beyond semiconductor companies. The more useful distinction is increasingly between assets that can generate enough cash flow to absorb a higher cost of capital and those whose return case depends heavily on financing remaining cheap.
Softer Inflation Does Not End the Higher-Rates Debate
PGIM’s 28 September Weekly View from the Desk argued that the forces pushing US borrowing costs higher are increasingly visible across other developed markets. Europe combines resilient economic activity with another energy shock, while governments must refinance debt at materially higher rates. PGIM expects further tightening from several central banks, including another 50 basis points from the Federal Reserve during 2026, while also forecasting increases from the European Central Bank, Bank of England and Bank of Japan. Those are PGIM’s forecasts rather than predetermined policy outcomes. [1]
The longer-term issue is fiscal arithmetic. PGIM identifies France and the UK as among the developed economies facing the most acute pressure as interest costs rise, and its accompanying chart shows government interest expenditure moving higher after years of exceptionally cheap funding. It therefore distinguishes between the possibility of a near-term bond-market consolidation and a more cautious long-run view of developed-market duration. [1]
BlackRock Investment Institute reached a compatible conclusion while differing on how much further monetary tightening markets should price. Its 28 September commentary reported that, during the preceding week, the 30-year US Treasury yield had reached 5.53%, its highest level since 2004, while the ten-year yield had risen to 5.22%. BlackRock believes expectations for further Fed tightening may have moved too far, but still prefers short- and medium-term Treasuries to long bonds because persistent inflation, high public debt and competition for capital make long duration a less dependable diversifier. [2]
Fresh inflation data added an important qualification on 30 September. The US headline PCE Price Index rose 3.4% year-on-year in August, unchanged from July’s revised reading, while core PCE was 3.0%, also unchanged from the revised July figure. Month-on-month, headline prices rose 0.3% and core prices 0.2%. The release incorporated the annual update of the National Economic Accounts, with revisions to personal income and outlays estimates beginning in January 2021. The figures remain well above the Fed’s 2% target, but the lower-than-expected release reduced market expectations for an immediate October increase. The revised annual readings should not be interpreted entirely as fresh disinflation in August. [5, 7]
That distinction matters. Softer inflation can alter the timing of the next policy move without resolving the forces influencing long-term yields. Government borrowing requirements, energy costs, AI-related capital demand and uncertainty over future inflation can all keep term premia elevated even if the Fed pauses at its next meeting.
Asian Bonds Have Been More Resilient Than the Global Headline Suggests
Eastspring Investments’ 25 September analysis provides a useful counterweight to the broad global bond sell-off. Over the previous month, two-, five- and ten-year US Treasury yields had increased by approximately 69, 66 and 50 basis points respectively. Most Asian ten-year government yields moved considerably less: around 23 basis points in India, 19 in Japan and Thailand, 17 in Singapore and Malaysia, nine in Indonesia and five in Korea, while China was broadly unchanged. [3]
Performance data show a similar divergence. As of 24 September, the J.P. Morgan Asia Credit Index was down approximately 0.3% year-to-date versus a 2.2% decline for the Bloomberg US Corporates Index. Asian local bonds, measured in US dollars, were down around 1.5% but remained ahead of US corporate credit despite the drag from a stronger dollar. Eastspring attributes part of the relative resilience to shorter duration, alongside credit spreads and market technicals that have held up comparatively well. [3]
For new capital, higher yields improve the prospective income proposition. Eastspring argues that the recent repricing has raised nominal yields, real rates and carry across the region. However, it explicitly rejects treating Asia as a homogeneous bond allocation. Its preference remains for quality carry at the shorter end and belly of curves while uncertainty remains elevated, with scope to extend duration once global conditions become more stable. [3]
This is relevant for private wealth portfolios because it separates two issues that are often conflated. Global rates can rise sharply while individual Asian markets respond differently because local inflation, policy credibility, current-account positions and currency conditions differ. A regional fixed-income allocation therefore offers potential diversification only if those underlying exposures are actually differentiated.
Credit Markets Are Still Absorbing Supply, but the Margin for Error Is Narrower
PGIM’s credit-market observations show that higher rates have not yet translated into a broad funding shutdown. Around USD38 billion of US investment-grade bonds were issued in the preceding week, with deals approximately 4.3 times oversubscribed and effectively no average new-issue concession. European investment-grade spreads also remained resilient despite widening sovereign spreads. [1]
Lower-quality credit experienced a more difficult week. PGIM recorded the worst weekly high-yield loss since April 2025 as the Treasury sell-off overwhelmed an earlier risk-on move. US high-yield issuance nevertheless reached approximately USD17.6 billion across 13 transactions, and investors were still buying weakness through exchange-traded funds. The market is therefore not displaying a wholesale withdrawal of capital, but the combination of rate volatility, higher oil prices and wider lower-quality spreads represents a meaningful deterioration from the previously benign backdrop. [1]
PGIM was more cautious on emerging-market hard-currency sovereign debt, reducing risk as spreads widened. It still sees selected carry opportunities, but describes the global rates bear market as increasingly entrenched and warns that no emerging market is completely insulated. Asia was comparatively resilient in foreign exchange during the week, while Latin American currencies experienced greater de-risking. [1]
The credit lesson is therefore similar to the rates lesson: headline yields are more attractive, but that does not make spread risk, refinancing risk or issuer selection less important. Higher carry provides a larger cushion, yet weaker companies now have to refinance into a substantially more expensive capital market.
AI Infrastructure Is Creating a New Financing Chain
BlackRock’s 29 September analysis of municipal bonds adds another layer to the AI financing story. It estimates AI-related municipal issuance could reach as much as USD11 billion in 2026, compared with projected total municipal issuance of roughly USD580–600 billion. The asset class is therefore unlikely to become the primary source of AI financing, but it can fund power networks, substations, transmission, water and wastewater systems supporting data-centre development. [4]
The wider scale is much larger. BlackRock cites JLL estimates that almost 100 gigawatts of global data-centre capacity could be added between 2026 and 2030, requiring up to USD3 trillion of investment. The financing chain spans corporate bonds, private credit, asset-backed and commercial mortgage-backed securities, project finance, municipal debt and equity. Different parts of the capital stack therefore carry very different exposure to the same underlying AI build-out. [4]
For municipal credit, the core question is not simply whether a community attracts a large technology company. It is who ultimately pays for the infrastructure. BlackRock highlights mechanisms including minimum-payment contracts, dedicated tariffs, upfront contributions, termination fees and parent guarantees. These structures can help prevent existing taxpayers or utility customers from carrying the cost of infrastructure built primarily for one private user. [4]
The mismatch between asset and financing life is also relevant. Computing hardware can become obsolete within several years, while municipal debt may amortise over decades. Long-lived grid or water infrastructure can therefore fit public-market financing much more naturally than rapidly depreciating servers and graphics processing units.
For wealth managers and family offices, this illustrates why an apparently diversified collection of municipal bonds, infrastructure funds, corporate credit, utilities and technology shares may still depend on the same capital-expenditure cycle.
Cybersecurity Emerges as an AI Adoption Trade Rather Than a Model Bet
Citi Wealth’s 29 September weekly update shifted attention from AI infrastructure to one of the spending categories created by wider adoption. It reported that its Global Investment Council, which met on 24 September, had increased exposure to cybersecurity alongside Japan and US large-cap equities, funding the change through a reduction in securitised fixed income. [6]
The logic rests on the growing security perimeter created by enterprise AI. Every additional AI agent can increase the number of identities, application connections and sensitive-data access points that companies need to authenticate and monitor. Citi cites Gartner forecasts showing security spending specifically for AI systems growing at a roughly 65% compound annual rate between 2026 and 2028, compared with approximately 13% for cybersecurity overall. [6]
That makes cybersecurity economically different from directly backing a model developer. A company does not necessarily need to know which foundation model ultimately dominates in order to conclude that greater AI adoption requires more protection of users, data and application programming interfaces.
The qualification is valuation. A durable secular spending theme is not automatically an attractive investment at every price. Citi’s argument is a portfolio view based on expected demand; security selection still requires assessment of revenue growth, margins, competitive position and the valuation already attached to those expectations.
China Is Moving Up the Industrial Value Chain, but Scale Is Not the Same as Return
BlackRock’s 28 September weekly commentary focused on China’s shift from low-cost manufacturing towards electric vehicles, batteries, advanced machinery and AI. China is not simply exporting more sophisticated goods; it is also increasingly supplying more of the inputs domestically. That creates benefits for buyers of cheaper technology while placing pressure on foreign producers competing directly with Chinese firms. [2]
BlackRock uses machinery as one illustration: China moved from a major customer of foreign advanced manufacturing equipment to a significant competitor, overtaking Germany as the world’s largest machine-tool exporter in 2025. In AI, cheaper Chinese models could similarly accelerate adoption while commoditising part of the model layer. [2]
The investment conclusion is deliberately selective. BlackRock remains neutral on Chinese equities overall because growing market share and industrial scale do not guarantee rising profit margins or stronger shareholder returns. It instead highlights areas where growth is translating into better economics, particularly physical AI and selected advanced-manufacturing exposures. [2]
That distinction is useful well beyond China. Industrial policy can create enormous output, lower global prices and reshape supply chains while producing very different investment outcomes for manufacturers, suppliers, competitors and customers.
Japan and US Large Caps Gain from Stronger Fundamentals
Citi’s equity changes provide another example of selective risk-taking rather than a simple shift towards or away from stocks. It moved Japanese equities from a slight underweight to a modest overweight, citing both stronger economic momentum and accelerating corporate earnings. Across MSCI Japan, earnings upgrades were running at approximately 2.6 times downgrades on a three-month basis, well above the longer-term average. [6]
Corporate profitability reinforces that signal. Citi reports Japanese profits at around 22% of output, a multi-decade high in its data, while investor positioning remains well below previous enthusiasm peaks. It also points to structural links with the AI supply chain, including Japanese industrial companies supplying materials used in high-bandwidth memory production. [6]
US large caps remain Citi’s main quality anchor. The S&P 500 led major regions in revisions to next-12-month earnings and carried a 9.4-times interest-coverage ratio in the analysis. Rising real yields have also compressed valuations, leaving the index’s price-to-earnings multiple around the 29th percentile of its range since 2020 while earnings growth remained in the 89th percentile. [6]
Citi funded the additions partly by reducing securitised fixed income to benchmark weight. Agency mortgage-backed securities remain high quality, but spreads of around 37 basis points were close to the tight end of their long-term range. Rising rate volatility also increases extension risk because higher mortgage rates slow prepayments and lengthen effective duration. [6]
The decision is therefore relative rather than absolute. Citi is not arguing that mortgage-backed securities have become poor-quality assets. It is arguing that expected compensation has become less compelling relative to equity opportunities where earnings and profitability are improving.
That is not a uniform view across the providers reviewed. PGIM was relatively positive on MBS in the short term, citing recent spread widening and dwindling supply, while BlackRock’s September positioning remained overweight US agency MBS. These assessments reflect different portfolio judgements rather than a shared decision to reduce the asset class. [1, 2]
What This Means for Wealth Managers and Family Offices
Hubbis draws the following editorial implications from the research reviewed.
Rate exposure should be decomposed rather than discussed simply as “bonds”. Government debt, investment-grade credit, high yield, Asian local rates and securitised assets are responding differently to the same global repricing. Shorter duration can offer substantial income without as much sensitivity to fiscal and term-premium risk, while longer duration requires greater conviction that inflation and capital demand will moderate.
Asian fixed income can provide differentiation, but the diversification comes from local fundamentals rather than geography alone. The much smaller yield moves across several Asian markets illustrate the value of independent inflation and policy cycles. That supports allocating by country, currency and curve rather than treating Asia as a single bond trade.
AI exposure needs a funding map as well as an earnings map. Data-centre operators, utilities, municipal issuers, private-credit vehicles, hyperscaler bonds and technology equities may all ultimately depend on continued AI capital expenditure. Looking through to the payer, contract structure, leverage and useful life of the underlying asset is increasingly important.
Second-order AI beneficiaries may offer a different risk profile from direct technology exposure. Cybersecurity, power networks, water infrastructure and advanced manufacturing can participate in AI adoption without relying on one model developer or application becoming dominant. That does not remove valuation risk, but it can diversify the source of earnings within the wider theme.
Equity broadening should be judged on profits rather than labels. Citi’s Japan allocation is based on stronger earnings revisions and profitability, while BlackRock’s China work shows why industrial growth alone is not sufficient. The same test applies to AI beneficiaries elsewhere: market share, investment and policy support need ultimately to become cash flow and returns on capital.
Liquidity remains a strategic asset in a high-rate environment. Attractive carry can tempt investors to deploy more capital into credit and private assets, but higher refinancing costs and market volatility increase the value of having cash available for capital calls, maturities and dislocations. The capacity to hold an asset through volatility can matter as much as the long-term investment thesis.
Hubbis’s broad conclusion from this week’s research is that higher rates have not broken the growth story, but they are forcing investors to become more precise about how that growth is financed. Asian bonds have shown relative resilience, Japan’s earnings backdrop has improved and AI continues to create new areas of demand. At the same time, governments, infrastructure projects and technology companies are all competing for capital in a world where money is no longer exceptionally cheap.
That favours portfolios built around distinguishable cash-flow drivers rather than broad thematic labels. The question is no longer simply whether an investor owns AI, Asia, equities or bonds. It is what sits beneath each exposure, who finances it, how sensitive it is to rates and whether the expected return adequately compensates for the capital required.
**
Sources
[1] PGIM, Global View on Debt, Inflation, and Rates, Weekly View from the Desk, 28 September 2026. Market returns table as of 25 September 2026. Original research
[2] BlackRock Investment Institute, weekly market commentary, 28 September 2026, and September 2026 tactical asset-class views. Weekly commentary | Asset-class views
[3] Eastspring Investments, Asian Fixed Income: Resilient through the rates shock, research note dated 25 September 2026. Supplied PDF; total-return data as of 24 September 2026.
[4] BlackRock, Municipal Bonds and AI Data Center Financing, 29 September 2026. Original analysis
[5] US Bureau of Economic Analysis, Personal Income and Outlays, August 2026, released 30 September 2026, and headline/core PCE data. Dated release | Headline PCE | Core PCE
[6] Citi Wealth, Following the Earnings: Where We Are Adding Equity Risk, Weekly Bulletin, 29 September 2026. Allocation decision dated 24 September 2026; key chart data as of 25 September 2026. Weekly update
[7] Reuters, US inflation rises below expectations in August, gives the Fed breathing space, 30 September 2026. Market reaction
**
Disclaimer
This article has been prepared by Hubbis for general information and editorial purposes only. It summarises selected third-party investment research published during the stated period and does not constitute investment, legal, tax or other professional advice, nor an offer, recommendation or solicitation to buy, sell or hold any investment or adopt any investment strategy. The views, forecasts and conclusions attributed to the research providers are those of the respective providers and do not necessarily reflect the views of Hubbis; the article also includes Hubbis’s editorial interpretation of the research. While reasonable care has been taken in preparing the summaries, Hubbis does not independently verify or guarantee the accuracy, completeness or continued relevance of the underlying information. Readers should review the original source materials and obtain appropriate professional advice before making any investment decision. Past performance is not indicative of future results, and all investments involve risk, including the possible loss of capital.
Leave a Reply