The macro backdrop has helped put a floor under the stock market tradeTakeawaysThe AI rebound looks more like a positioning reset than a clean all-clear. The next verdict comes from earnings, cash flow and capex guidance.Softer jobs data and lower oil have taken some heat out of the Fed-hike trade, reopening the liquidity bid across Nasdaq, gold and Bitcoin.Gold and Bitcoin have slipped back into the same macro bucket: less rate-hike risk, a softer dollar and easier financial conditions.USD/JPY remains the tail-risk trade. Thin holiday liquidity and a less predictable intervention threat make chasing dollar strength above 161 an expensive habit.The Return of the Liquidity Trade?With US cash markets shut for the July 4 holiday, futures have found their footing after the latest AI-led momentum washout, although thin liquidity is doing its usual job of making every move look a little larger than it probably is. S&P futures are up modestly while Nasdaq futures have bounced more sharply, helped by a recovery in South Korean memory names after two bruising sessions. SK Hynix and Samsung have led the rebound, with the broader Asian complex clawing back early losses as investors decided that a two-day air pocket in semiconductors was not yet the same thing as a broken AI cycle.That distinction matters. The selloff had started to look less like a sober reassessment of earnings and more like a crowded momentum trade discovering it had too many passengers leaning over the same side of the boat. SK Hynix had fallen roughly 30% from its recent peak at one stage, which was enough to trigger the usual reflexive unwind across chips, AI infrastructure, and the leveraged products tied to them. Once the selling becomes mechanical, the price action often runs ahead of the underlying story. Friday’s rebound is therefore not proof that the valuation argument has disappeared, but it does show that the market was becoming stretched on the downside.The next real test comes with earnings. The central question has not changed: can the extraordinary AI infrastructure spend translate into a durable profit pool rather than simply a larger capex bill? The bullish case remains that memory, networking, power, and compute suppliers are still positioned for a multi-year investment cycle. The common theme from the bullish side of the street is that profit expectations across memory and the wider AI hardware supply chain may still be underpriced. The bears will counter that the market has already priced in much of that future, and that is precisely why every earnings update now carries more weight than the last.The macro backdrop has helped put a floor under the stock market trade. Softer US payrolls, falling oil prices, and a more cautious inflation read-through have taken some heat out of the Fed-tightening story. The first fully priced 25 basis-point hike has shifted back toward December from October, and that repricing has weakened the US dollar while lifting duration-sensitive and liquidity-sensitive assets. Gold has pushed toward $4,170 an ounce, its strongest level in nearly two weeks, while Bitcoin has bounced back above $62,000 after its own recent unwind.Gold and Bitcoin are not the same asset, but in this kind of market, they can end up trading out of the same liquidity bucket. When the market begins to mark down the odds of higher real rates and a stronger dollar, both can catch a bid from the same change in the funding weather. Their recent dislocation was a sign that positioning had become uneven. Their rebound suggests that the macro overlay is reasserting itself, at least for now.The dollar has felt the other side of that adjustment, touching a two-week low as rate-hike expectations soften. USD/JPY has also been volatile again, including another air pocket through 161 that will keep intervention risk high on traders’ screens. The yen does not need Tokyo to be active every day to make the market nervous; uncertainty over the timing and method of official action is often enough to keep leveraged dollar longs from getting too comfortable above 161.Asia remains the clearest expression of the AI concentration trade. Samsung, SK Hynix, Kioxia and TSMC account for a meaningful share of the regional benchmark, so the index effectively carries a large embedded view of the global compute cycle. Friday’s rebound was helped by reports that Anthropic is in discussions with Samsung about custom AI chip manufacturing, while the wider rally also reflected relief from lower oil prices and softer US employment data. But the concentration cuts both ways. When a handful of chip names wobble, Asia’s benchmark can suddenly behave like a very expensive semiconductor ETF.Europe, meanwhile, has continued to grind higher, with the Stoxx 600 near record territory as utilities and technology provide the lift. The contrast is notable. US investors have pulled $17.2 billion from domestic equity funds in the week through July 1, according to Bank of America’s flow data, while Japanese equities drew their largest inflow in seven weeks. That does not mean the US AI trade is finished. It does suggest that investors are becoming more selective about where they want to pay peak multiples, especially with international markets offering a cheaper way to express a softer-dollar, easier-Fed view.For traders, the immediate question is whether this is a genuine reset or just a holiday-thin bounce after forced selling. The answer will come quickly. Samsung’s preliminary earnings on July 7, SK Hynix’s Nasdaq ADR listing next Friday, and the next batch of AI-related guidance will determine whether the memory complex can rebuild credibility. Until then, the rebound should be respected, but not romanticized. The AI trade has not lost its growth engine. It has simply been reminded that when valuations, leverage and momentum all crowd into the same room, the exit can get narrow very quickly.Snippets from the most popular reads behind this week’s paywallEarnings Season Is About to Decide Whether the AI Rally Has Another LegThe next real test for US equities arrives with second-quarter earnings in the week of July 13. According to Goldman Sachs Research strategist Ben Snider, the market’s 21% return over the past year has been entirely earnings-driven, which leaves very little room for disappointment now that valuations are already carrying a full load.That puts the upcoming reporting season at the centre of the market’s next move. Consensus expects S&P 500 earnings per share to rise 22% year on year in the second quarter, while Goldman forecasts 24% EPS growth for 2026 as a whole. Those are not soft numbers. They imply that companies need to keep delivering, not merely avoid a miss.The pressure point will be the large technology and hyperscaler complex. Investors will be looking less at the headline revenue print and more at whether the enormous AI capital-spending cycle is starting to produce an economic return. The first quarter offered some encouragement: revenue estimates moved higher, backlogs expanded and gross margins improved. That gave the market enough evidence to believe the compute buildout was beginning to convert into actual monetisation rather than simply bigger capex budgets.But the market has become less patient in recent weeks. Weak free cash flow, heavier equity and debt issuance, and uncertainty over who ultimately captures the economics of frontier models have started to weigh on hyperscaler shares. The issue is no longer whether AI demand exists. It plainly does. The issue is whether the companies funding the infrastructure build are the same ones that will earn the most from it.That is why this earnings season matters more than usual. The AI trade is moving from story stock territory into proof-of-return territory. Strong earnings and credible guidance could put the recent momentum wobble behind it. But if capex keeps rising faster than cash generation, investors may begin to treat the infrastructure race less like a gold rush and more like a very expensive arms race.AI Is Turning Into a Capital Markets TradeThe AI boom is no longer just a race for chips, models and data centres. It is becoming a race to finance them.Goldman Sachs Investment Banking estimates the AI buildout could require roughly $7 trillion between 2026 and 2031, a number that pushes the story well beyond the usual hyperscaler capex cycle. At that scale, equity markets, public debt, private credit, sovereign wealth funds, pension capital and joint-venture structures all need to show up. The financing stack is getting wider because the bill is getting too large for any one corner of the market to carry.That is the deeper shift now underway. AI is beginning to reorganise the industrial economy, not just the technology sector. Software was the first visible casualty, with investors repricing the threat of AI disruption across the sector earlier this year. But manufacturing, robotics, defence and energy are still only at the opening stages of the same process. The winners will not simply be the firms with the best technology. They will be the ones able to secure enough capital, power and infrastructure before the cycle moves on.The funding numbers are already moving at a pace that should make investors pay attention. Hyperscale companies had issued around $107 billion of debt by June 23, already above 2025’s full-year total and far larger than the less than $20 billion raised in 2024. That is a major shift in how the AI buildout is being financed. What started as an equity-market story backed by huge operating cash flows is increasingly becoming a credit-market story as well.Private capital is moving into the gap. Infrastructure funds raised a record $221 billion last year and could expand toward roughly $3 trillion in assets by 2030, according to Goldman’s analysis. That is not accidental. Traditional bank lending and public capital markets were not designed to finance a wave of giant, power-intensive data campuses, grid upgrades and bespoke compute infrastructure all at once.The market is now inventing new ways to do it. Private credit is financing individual data-centre campuses above one gigawatt. Sovereign wealth funds and pension funds are shifting from passive allocators to direct co-investors. Leveraged finance and high-yield markets are funding infrastructure operators that do not fit neatly into investment-grade credit boxes. The AI trade is becoming more institutional, more structured and, inevitably, more levered.That is where the opportunity sits, but it is also where the fault line starts to form. A broader capital base can extend the AI investment cycle far beyond what public equity investors might tolerate on their own. But as the buildout migrates into private credit, high yield and structured financing, the question changes. It is no longer only about whether AI demand keeps growing. It becomes whether the returns arrive quickly enough to justify the increasingly complex capital being used to fund it.The next phase of AI may not be won by the company with the best model. It may be won by the company that can keep funding the machine when everyone else starts running short of balance sheet.Why Gold Could Find Its Footing Again in the Second HalfGold has had a rougher first half than many expected. After rallying 123% since 2022, the metal gave back roughly $300 in the first six months of 2026, falling to around $4,016 an ounce by late June as markets began pricing a more hawkish Federal Reserve path and the dollar found some renewed traction.But Goldman Sachs Research argues that the setback may prove more like a pause than the end of the move. Its year-end target sits at $4,900, built around three familiar but powerful pillars: continued central-bank buying, a later shift in rate expectations from hikes toward cuts, and private investors returning to gold as a diversification trade.The central-bank story remains the structural backbone. Buying has slowed recently, but the broader reserve-diversification impulse has not disappeared. A World Gold Council survey found that 45% of the 76 central banks surveyed between February and May expect to increase gold holdings over the next 12 months, the highest reading in the survey’s history. For emerging-market reserve managers in particular, gold remains one of the few assets that can reduce dollar concentration without simply adding another sovereign credit exposure.The near-term problem has been the Fed. Inflation worries and higher oil had pushed markets toward pricing rate hikes this year, which is never an easy backdrop for gold. Higher expected rates raise the opportunity cost of holding a non-yielding asset, while a firmer dollar can take some of the wind out of the trade. Rate-sensitive ETF demand has also softened as investors began to question whether the easing cycle was further away than previously assumed.That is the key hinge for the second half. Goldman’s view is that the Fed ultimately cuts rather than hikes, although not until next year. The timing matters less than the direction of travel. Once markets start moving from pricing further tightening to pricing eventual easing, gold tends to find a more supportive macro current. The metal does not need rates to collapse. It simply needs the real-yield and dollar headwind to stop getting stronger.Gold’s recent decline has therefore been less about a collapse in the long-term bull case and more about the market temporarily repricing the policy weather. Central banks are still diversifying, private portfolios remain heavily exposed to financial assets, and the rate market may yet have to unwind some of its hawkish excess.For traders, the important point is that gold is once again behaving like a rates and dollar proxy. The next leg higher will likely not come from a dramatic headline. It will come when the market decides that the Fed’s tightening threat has run out of runway and the reserve-diversification bid is still quietly sitting underneath the price.Running UpdateAfter eight heavy weeks, the body finally tapped the brakes this morning. Even though it was only a 7 km easy run, the legs felt unusually heavy after lifting the pace across the previous four outings, which is usually my signal that a proper 48-hour reset is due.No drama, just accumulated fatigue doing what accumulated fatigue does. Better to listen now than turn a good training block into an unnecessary niggle. The timing is perfect for a quiet, restorative weekend in Hua Hin: sleep, easy walking, decent food, and no need to prove anything to Dr Garmin.