BofA Securities chief strategist Michael Hartnett has issued a clear warning: a further decline in the ETF tracking the Mag7 tech giants, combined with a stronger yen and an inverted yield curve, could trigger a broad market shift into defensive mode this summer. The structural divergence where hyperscale cloud providers continue to underperform chip stocks is pushing the sustainability of AI capital expenditure to the forefront of current market debates.

In BofA's latest Flow Show report, Hartnett explicitly lists three thresholds that could trigger a "full defensive summer": the Mag7 ETF (ticker MAGS) falling below $60, the USD/JPY rate dropping below 110, and the yield curve inverting again. His central question targets the market's most sensitive nerve: "How much do hyperscalers need to fall before the market starts trading capex cuts?" This has been his recurring theme over the past month.

Signs of a turning point are already visible in fund flows. U.S. equity funds recorded $8.5 billion in net outflows, the first such instance since March 2026, following a historic $119.2 billion net inflow record for the asset class. Simultaneously, capital flowing out of the tech giants is accelerating into more cyclical, less liquid assets like semiconductors, small and mid-cap stocks, housing, and REITs.

On a broader macro level, Hartnett views the U.S. dollar as a "rent" (short-term hold) rather than an "own" (long-term allocation) asset. He sees strong allocation value in gold below $4,000, while strategically advocating for long positions in emerging markets and long-dated U.S. Treasuries.

Three Thresholds Define the Defensive Tipping Point

Hartnett identifies three key indicators as potential triggers for a full market shift to defensive positioning: the Mag7 ETF falling below $60, USD/JPY dropping below 110, and the yield curve inverting. None of these conditions have been met yet, but sustained pressure on hyperscalers has significantly heightened market vigilance.

Currently, corporate profit margins remain sticky at a high 16%, providing some support for the overall preference for equities. Hartnett notes that liquidity exiting the tech giants is pre-positioning in cyclical assets like semiconductors, small/mid-caps, housing, and REITs. The market interprets this as front-running an expected policy shift towards "affordability" under a potential Trump administration—though such a shift remains uncertain.

Hyperscalers Underperform as Chip Stocks Lead

The divergence between hyperscale cloud providers and chip stocks is now the most prominent structural feature within the tech sector. The former, as the payers for AI infrastructure capex, face continued stock price pressure, while the latter, as beneficiaries of that spending, show strong gains.

According to Rich Privorotsky, head of Goldman Sachs' Delta One business, in his latest report, hardware beneficiaries continue to strengthen, while hyperscalers and the memory consumer side are clearly underperforming the broader market.

Recent price increases for some MacBook lines by Apple and for Xbox hardware by Microsoft, both closely linked to rising memory costs, confirm the logic that AI infrastructure costs are being passed down the supply chain. Privorotsky writes, "Aggressive margin capture in memory and HBM supply shortages have caused a significant transfer of value from consumers to producers."

Data shows memory prices in Vera Rubin racks have surged 435% cumulatively, directly squeezing returns on capital for hyperscalers. Rumors of Apple seeking to source memory chips from China are also seen as an active response to high costs.

The Capex Debate: Shareholder Pressure Emerges

The scale and rationale of AI infrastructure capital expenditure has become one of the market's core debates. Goldman Sachs projects this figure could reach as high as $1.4 trillion by 2027.

Privorotsky points out that markets have historically rarely rewarded companies that consume large amounts of free cash flow for capex during the build-out phase. "This doesn't mean management's long-term decisions are wrong, just that equity investors typically prefer immediate, tangible returns over distant, uncertain cash flows. Valuation expansion usually occurs during the harvest phase, not the build phase." He further notes that if hyperscalers continue to underperform while suppliers keep gaining, corporate boards may increasingly question whether incremental AI investments are truly maximizing shareholder value—"once one peer wavers, the market will immediately question whether others should follow."

Hartnett's framework aligns closely with this logic. In his view, the key question now is precisely: what is the tipping point at which hyperscalers are forced to constrain capital expenditure?

Fund Flows Show First Net Outflow

U.S. equity funds recorded $8.5 billion in net outflows, the first instance since March 2026. This reversal follows a historic $119.2 billion net inflow record, which Hartnett sees as a signal that equities may have reached a local unfavorable inflection point.

While a single week's data is insufficient to confirm a trend reversal, combined with sustained pressure on the Mag7 and the widening divergence between hyperscalers and chip stocks, the market's internal structure is quietly shifting.

Bonds Outperform Since Wash's Inauguration, Long-Dated Treasuries a Contrarian Pick

Since Federal Reserve Chair Wash officially took office on May 22, U.S. Treasuries have gained 3.2% cumulatively, while stocks have fallen 1.6%, showing clear relative outperformance by bonds.

Hartnett believes that, although it's early and the new chair's hawkish tilt hasn't yet prompted investors to abandon the "Anything But Bonds" core allocation logic, Wash's market impact so far is effectively mirroring the paths of historically yield-lowering Fed chairs like Eccles, Volcker, Greenspan, and Bernanke. Based on this, Hartnett views long positions in long-dated U.S. Treasuries as the most contrarian long-term trade in the market currently.

Gold Holds Allocation Value, Dollar for Trading Only

On major asset classes, Hartnett holds a clear strategic bullish view on gold: below $4,000 remains an excellent buying opportunity. The U.S. dollar, in contrast, is merely a "rent" (short-term hold) asset, not one for long-term "ownership."

He believes the 2020s will remain an era dominated by geopolitical fragmentation and populist politics, with policy focus continuing to tilt towards economic prosperity over inflation control. In this context, a long-term long position in emerging markets is one of his core strategic calls. For silver, Hartnett identifies below $60 as a reference for a stress scenario. Bitcoin below $60,000 is also incorporated into a "new normal" indicator system. He notes that potential near-term dollar strength from a possible Iran deal and the end of the sanctions era is a source of potential pressure for these assets.