Macro
The US data mix turned materially less hawkish. July payrolls fell 23k vs +80k expected, compounded by 103k of downward revisions to May/June; unemployment nevertheless fell to 4.1% vs 4.2% expected, partly reflecting weaker participation, while initial claims rose to 209k. Inflation simultaneously moderated as the energy shock faded: headline CPI slowed to 3.4% YoY from 3.5%, core CPI to 2.5% from 2.6%, with core +0.2% MoM, matching its slowest annual pace since March 2021. PPI was flat MoM vs +0.2% expected, taking the annual rate to 4.7% from 5.5%; core PPI ex-food, energy and trade was firmer at +0.4% MoM but slowed to 4.7% YoY from 5.0%. Pipeline indicators also suggest input-price pressures are cresting. The combination of weaker employment and improving inflation strengthens the case for the Fed to remain at 3.75% through 2026, although Cleveland Fed’s Hammack continues to flag possible hikes and Deutsche Bank remains a notable outlier forecasting 25bp increases in September and December.
Importantly, weaker payrolls have not yet translated into broad economic contraction. NFIB small-business optimism rose from 97.4 to 99.8, near a one-year high, with hiring intentions reaching their strongest since October 2022 and planned capex the highest since late 2024. Consensus therefore continues to imply slowdown rather than recession: Bloomberg has 2026 GDP growth at 2.16%, slowing from 2.71% YoY in Q1 and 2.20% in Q2 to 1.67% in Q3, before recovering to 2.04% in Q4; institutional forecasts range from 1.9% at Fitch to 2.3% at the IMF/ING, with recent 12-month recession estimates around 20–25%. AI capex remains the major offset to labour weakness, with hyperscaler infrastructure spending potentially exceeding $750bn in 2026, +82% YoY. Consumer fundamentals are less reassuring: preliminary Michigan sentiment fell to 51 vs 55 expected, one-year inflation expectations rose to 4.3%, savings have declined, and credit-card delinquencies remain elevated.
Fiscal deterioration remains a structural constraint. Fitch affirmed the US at AA+/stable but expects the general-government deficit at 7.4% of GDP in both 2026 and 2027, with debt rising from 117% of GDP in 2025 to 123% by 2028, driven by tax cuts, tariff rebates, defence and entitlement spending and higher interest costs. A record July budget deficit reinforced the trajectory. Large structural deficits leave less capacity for countercyclical support if growth deteriorates and are increasingly a common constraint across developed economies.
The Middle East remains the largest macro tail risk, principally through energy. There was no meaningful progress toward resolving the US-Iran conflict or restoring normal traffic through the Strait of Hormuz, while Houthi activity around Bab al-Mandab adds a second shipping chokepoint. The IEA estimates a 1.8mn b/d Q3 oil deficit, the largest in five years; WTI rebounded roughly 6% to $81.84/bbl and Brent 5.8% to $87.47/bbl, despite reported US crude inventory builds of 9.1mn barrels (API) and 17.4mn (EIA). At current prices, the shock remains manageable, but the tail is asymmetric: Morgan Stanley estimates sustained $120/bbl oil would materially slow growth and keep inflation elevated, while $140–160/bbl through Q3 could generate two consecutive quarters of negative US GDP growth and push headline CPI toward 5.1%. Oil therefore remains the key variable that could reverse the recent improvement in US inflation.
Trade fragmentation adds a second supply-side risk. New US tariffs on Switzerland replaced the previous 10% regime with additional tariffs of up to 12.5%, while proposed legislation could enable tariffs of at least 100% on major buyers of Russian oil, including China, India and Turkey. Potential USMCA revisions are also raising costs for North American automakers. Meanwhile, China’s accelerated removal of foreign technology from state institutions reinforces the broader US-China move toward duplicated technology ecosystems. Tariffs, energy insecurity, and technology decoupling collectively imply a structurally less efficient (and potentially more inflationary) global supply environment.
Outside the US, growth remains resilient, but policy divergence is widening. Eurozone Q2 GDP grew 0.4% QoQ/1.0% YoY, while France’s central bank expects momentum to continue in Q3; however, German inflation accelerated to 2.8% from 2.4% and Spanish inflation to 3.9% from 3.6%, limiting ECB flexibility. The UK expanded 0.3% MoM in June and 0.4% QoQ in Q2, taking Q2 growth to 1.2% YoY, although industrial production (-0.2%), manufacturing (-0.5%), weaker retail growth and a wider £5.54bn trade deficit point to softer underlying momentum.
Asia presents the opposite inflation mix. China CPI slowed to 0.5% YoY from 1.0% and fell 0.1% MoM, while producer-price inflation eased for the first time since the Iran war began as the oil shock faded and domestic demand remained weak. Japan’s economy is firmer, with Economy Watchers improving to 45.7 from 44.0 and Reuters Tankan to 18 from 13; importantly, the government is now reportedly supportive of a near-term BOJ hike, potentially September or October, as yen weakness continues to threaten domestic inflation. Australia similarly remains hawkish: the RBA held at 4.35% for a second meeting, with Governor Bullock warning further tightening remains possible. The global policy backdrop is therefore increasingly fragmented: the US is moving toward an extended pause as labour weakens and inflation cools, while several economies facing stickier inflation retain a tightening bias.
The base case remains slower but positive US growth, continued underlying disinflation and an extended Fed pause through 2026. The combination of weak payrolls and softer CPI/PPI suggests peak Fed hawkishness has probably passed, while improving small-business hiring and capex intentions and extraordinary AI investment argue against interpreting the labour shock as an imminent recession.
The distribution remains unusually wide, however. AI capex is the principal growth support; oil is the principal macro risk. If the Hormuz shock continues to fade, inflation should moderate further, and the soft-landing case strengthens. If crude moves sustainably above $120–140/bbl, the combination of weaker labour, fragile consumer confidence and renewed inflation would shift the regime rapidly toward stagflation and leave the Fed with very limited room to respond.
Rates
U.S. Treasuries steepened sharply over the week as weaker labour data and benign inflation reduced near-term Fed tightening expectations, while fiscal, supply and term-premium pressures continued to weigh on longer maturities. The 2Y yield fell 2.4 bps to 4.173%, while the 5Y rose 1.2 bps to 4.366%, the 10Y rose 4.7 bps to 4.694%, and the 30Y rose 6.0 bps to 5.263%. The result was a pronounced twist-steepening of the curve: 2s10s widened 7.1 bps to 52.1 bps, 2s30s 8.3 bps to 108.9 bps and 5s30s 4.7 bps to 89.7 bps.
The front-end rally was driven primarily by the July payroll report, where employment fell 23k versus expectations for an 80k increase, a 103k downside surprise. CPI subsequently offered little reason for renewed hawkishness, with headline inflation at 3.4% YoY and core at 2.5%, both in line with consensus. Markets responded by cutting the probability of a September Fed hike from 43.1% to 31.9%, while the implied December policy rate declined 4.4 bps to 3.864%. Cleveland Fed President Beth Hammack nevertheless maintained a hawkish stance, arguing that inflation has remained above target for too long and that policy should be tightened further.
The more important signal came from the long end. Despite declining Fed hike expectations, 10Y and 30Y yields moved higher, increasingly separating the outlook for monetary policy from the compensation investors require for holding duration. The 10Y auction cleared at 4.683%, around the highest yield since 2007, but demand was firm: it stopped 0.1 bp through the when-issued level, with a 2.53x bid-to-cover and 76.7% indirect participation. This suggests elevated yields reflected the prevailing market-clearing level rather than weak auction demand. The 30Y remained above 5.25%, with recent issuance clearing near the highest yields in roughly 25 years.
Inflation pricing reinforces the view that the long-end selloff was not primarily an inflation-expectations shock. 5Y and 10Y breakevens increased only 2.3 bps and 3.0 bps, ending at 2.253% and 2.284%, respectively. By contrast, real yields rose further out the curve: the 10Y TIPS yield increased 1.9 bps to 2.414%, and the 30Y climbed 4.7 bps to 3.033%, crossing 3% during the week. The 5Y real yield actually fell 1.1 bp. The divergence points toward higher real rates and term premium (rather than materially unanchored inflation expectations) as the primary source of long-end pressure.
Fiscal and supply concerns remain central to that repricing. Fitch maintained the U.S. sovereign rating at AA+ with a stable outlook, while projecting general government deficits around 7.4% of GDP in both 2026 and 2027 and debt rising toward 123% of GDP by 2028. Treasury supply is also increasingly competing with substantial corporate duration issuance, particularly from technology hyperscalers funding AI-related capex. Tech has accounted for roughly 20% of U.S. investment-grade issuance YTD, adding another potential source of upward pressure on longer-dated yields.
Despite the long-end volatility, current yield levels continue to improve the strategic case for high-quality fixed income. Real yields around or above 3% at the long end and nominal yields near 5% provide substantially greater carry protection than in the post-GFC low-rate regime. The 7–10Y belly remains the cleaner relative-value expression, offering attractive carry and roll-down while limiting exposure to the fiscal and supply risks concentrated further out the curve. Strong inflows into long-duration Treasury products nevertheless suggest investors are increasingly viewing current long-end yields as attractive entry levels.
Japan added to the global duration pressure. The 10Y JGB yield rose 8.7 bps to 2.892%, and the 30Y increased 10.0 bps to 4.031% as markets priced a greater probability of additional BoJ tightening. With September hike expectations elevated, further Japanese normalisation remains an important potential spillover channel into global yields, particularly if higher domestic yields encourage Japanese investors to repatriate capital from overseas bond markets.
Overall, Week 33 reinforced a two-speed rates environment: weaker labour data and contained inflation reduced near-term Fed tightening risk and supported the front end, while fiscal concerns, heavy supply and rising real term premium pushed longer yields higher. The key implication is that easier Fed expectations do not automatically translate into lower 10Y and 30Y yields. Unless fiscal expectations improve or term premium compresses materially, the Treasury curve can continue to steepen even against a relatively benign inflation backdrop.
Credit
Credit remained technically strong, but the scale and structure of AI infrastructure financing is becoming a key medium-term spread risk. Nvidia announced a coalition of major investment firms targeting >$500bn of financing, against an estimated $5tn–$7tn required for the broader AI buildout. Nvidia CDS tightened after the announcement, but key uncertainties remain about how much financing ultimately takes the form of debt, increasingly circular financing structures, and the residual value of AI chips as newer, cheaper generations replace existing hardware. Nvidia is reportedly providing guarantees of sorts to mitigate some of these concerns, while insurance companies and private-credit lenders are expected to absorb substantial amounts of the financing.
Primary supply remained exceptionally heavy: >$50bn of high-grade bonds priced during the week, while August issuance had already reached ~$130bn by mid-month, effectively meeting the full-month forecast. Blue Owl issued $750mn to repay credit-facility borrowings, while AMD also returned to market. Goldman Sachs expects the U.S. bond market not to carry the AI financing requirement alone: funding should increasingly span public IG, private credit, structured finance, multiple currencies and off-balance-sheet structures, with activity remaining elevated into 2027.
The bigger issue is rising supply concentration and pricing, not immediate fundamental deterioration. Hyperscalers have accounted for roughly 30% of net IG issuance YTD, with relative underperformance already emerging as issuers have levered up. Investors expect at least another $500bn of hyperscaler-related issuance over the next 12–18 months, spanning on-balance-sheet bonds, off-balance-sheet financing, private credit, HY and leveraged loans. Investors therefore see little urgency to overweight hyperscalers, expecting continued supply and some indigestion to widen spreads and create better entry points. With AI exposure now distributed across virtually every credit asset class, investors also need to aggregate exposures across structures and identify the ultimate guarantor to avoid unintentionally owning the same underlying risk multiple times.
Recent large technology financings also suggest that long-term buy-and-hold investors may ultimately demand cheaper debt as supply increases; the discussion highlighted some SpaceX debt tranches falling by roughly 10 points. Future financing from other major AI companies could further test market capacity, reinforcing the likelihood of repricing as issuance expands.
High yield remains expensive at ~280bp, providing limited compensation for volatility or credit losses relative to the past 25 years. Beneath the headline index, however, CCCs (~10% of HY) have materially underperformed over recent months, with stress concentrated particularly in software and cable, alongside several idiosyncratic credits. This is partly a healthy sign that markets are discriminating against weak businesses rather than indiscriminately providing capital, but it also indicates growing vulnerabilities beneath otherwise tight aggregate spreads.
Overall, the strong technicals continue to support headline spreads, but unprecedented AI financing needs are creating a structural supply and concentration challenge. With hyperscalers already representing ~30% of net IG issuance YTD and another ≥$500bn potentially coming over 12–18 months, patience and cross-asset exposure management are increasingly important. Meanwhile, ~270bp HY spreads conceal meaningful CCC stress, reinforcing the case for issuer selection over broad credit beta.
Equities
Global equities posted a mixed week, with Japan leading developed markets and Energy dominating S&P 500 sector performance. The Nikkei 225 was the standout, surging 4.74% (or 3.71% in USD), while MSCI EM gained 2.61%. Within the US, the equal-weight S&P 500 outpaced the cap-weighted index, rising 1.15% versus 0.36%, a 79 bp spread in favour of equal-weight and a clear sign that breadth, rather than mega-cap concentration, drove the modest US advance. The Russell 2000 added 1.12%, broadly in line with equal-weight and reinforcing the small- and mid-cap tilt to the week’s gains. The Nasdaq eked out just 0.14%, while the Dow slipped 0.56%. Outside the US, Germany’s DAX gained 0.55%, while the FTSE 100 fell 1.07% and the Hang Seng declined 2.17%.
Energy (+7.31%) dominated S&P 500 sector performance by a wide margin, supported by higher oil prices and renewed Middle East risk. Defensives also held up relatively well, with Utilities (+1.45%), Consumer Staples (+0.99%) and Health Care (+0.98%) all advancing, suggesting a degree of risk-off rotation alongside the energy move. Financials (+0.90%), Industrials (+0.68%) and Real Estate (+0.66%) posted more modest gains. Information Technology (+0.22%) was nearly flat, consistent with the Nasdaq’s muted performance, while Materials (−0.87%), Communication Services (−0.96%) and Consumer Discretionary (−1.97%) were the week’s laggards.
The broader message from the US tape was one of rotation rather than outright risk aversion. Value continued to improve relative to momentum, while equal-weight and small caps outperformed. Fundamentals also supported the broadening story, but the headline ~32% y/y S&P 500 Q2 EPS growth overstates the underlying pace because of a large one-time investment gain at Alphabet. Excluding Alphabet, earnings growth is closer to ~25.9%, still exceptionally strong. More importantly, the broadening is real: Mega-cap Growth & Tech now contribute roughly 55–57% of aggregate S&P 500 earnings growth, down from around 90% a year ago; eight of 11 sectors are delivering double-digit EPS growth, and 19 of 24 industry groups have seen upward revisions. Small caps are also on track for roughly 11% revenue growth in Q2, ahead of pre-season expectations, reinforcing the view that earnings strength is becoming materially less concentrated in the largest technology names.
The AI trade remained the market’s dominant structural theme, but the debate is increasingly shifting from AI exposure to AI economics. Technology experienced meaningful intra-week volatility, with Nvidia and Intel weakening earlier in the week and Cisco selling off sharply despite a headline earnings beat, after its AI-specific revenue guidance disappointed. The reaction illustrated how high the bar has become for AI monetisation. At the same time, the top 10 S&P 500 stocks now account for 39.2% of index market capitalisation, leaving the market sensitive to any deterioration in AI earnings expectations or valuation support.
More fundamentally, the focus is moving toward which parts of the AI ecosystem can convert extraordinary capex into sustainable returns. BCA’s work argues that hyperscalers are structurally better positioned than NeoClouds and standalone frontier-model providers because they can monetise AI infrastructure across existing cloud, software and platform ecosystems. NeoCloud economics remain attractive while compute is scarce, but returns may compress as capacity normalises, while frontier LLM providers face the additional risk of model commoditisation, price competition and vertical integration by hyperscalers. The emerging hierarchy therefore favours platforms and hyperscalers as prospective AI monetisers, followed by semiconductors and infrastructure beneficiaries whose economics remain more dependent on continued capex growth.
Corporate dispersion was unusually high, with the largest moves concentrated in Energy, memory and AI infrastructure, and event-driven names. Marathon Petroleum (+19.2%), Phillips 66 (+14.6%) and Valero (+14.5%) surged as elevated refining margins extended the Energy rally, while Phillips 66 also announced a $5bn Western Gateway Pipeline JV. The standout was Sandisk (+35.4%), whose investor day outlined a 2028 to 2030 model targeting mid-to-high-teens annual revenue growth, around 80% non-GAAP gross margins, 75% operating margins and 50% adjusted FCF margins, reigniting enthusiasm across the memory complex. Seagate (+19.8%), Western Digital (+17.2%), Micron (+10.7%) and Super Micro (+28.0%) rallied in sympathy. Workday (+10.6%) jumped on reports that Silver Lake was exploring an acquisition, while Intel completed a heavily oversubscribed $20bn equity raise, its first public share sale since 1971, strengthening confidence around its foundry build-out despite dilution.
On the downside, Cisco (−8.0%) highlighted the market’s rising expectations for AI delivery. The company beat on headline earnings and revenue and guided FY2027 above consensus, but its $7.5bn AI revenue outlook disappointed relative to the $9.3bn AI-related order backlog built over the prior year. Tapestry (−20.6%) sold off on softer FY2027 guidance and a slower Kate Spade recovery, while Broadcom (−8.1%) and AppLovin (−9.0%) were hit by profit-taking and lingering concerns around high-flyer positioning. Elsewhere, PayPal gained on renewed takeover speculation, while Reddit rallied after being selected for S&P 500 inclusion.
Positioning remains supportive, but increasingly crowded. Fund-manager equity allocations are elevated, and cash balances are low, while hedge funds recorded their strongest net equity buying in six months. At the same time, speculative investors increased Nasdaq 100 net shorts to their highest level of 2026, showing that broader bullishness is being paired with more caution toward large-cap technology. The AI capex cycle remains central to the outlook: Alphabet, Microsoft, Meta and AWS spent $165.6bn in Q2, up 108% y/y, with consensus projecting combined 2026 capex of roughly $711bn, up 88% y/y. Nearly two-thirds of surveyed investors reportedly want clearer evidence of AI monetisation before increasing technology exposure further, while 59% are hedging AI downside by rotating into value, cyclical and defensive sectors.
This week we’ve seen changing characteristics of the rally. Earnings growth is broadening, equal-weight and small caps are outperforming, value is gaining relative to momentum, and Energy has re-emerged as a major source of leadership. At the same time, investors are becoming more discriminating within AI, separating companies simply exposed to infrastructure spending from those that can convert that investment into durable free cash flow and higher returns on capital. The backdrop remains constructive, but with positioning increasingly bullish and the bar for AI monetisation rising, future upside is likely to depend more on earnings delivery, breadth and capital efficiency than further multiple expansion.