Google's $80 billion equity issuance broke the correlation between hyperscalers and memory stocks
For months, DRAM and the Mag 7 moved in lockstep. Then Google sold $80 billion in equity, and the relationship inverted. The rotation that followed has split two corners of the chip market that investors had been treating as one trade.
A correlation that held across months of AI-driven capital flows broke sharply in early June. Jack Farley put it plainly: the positive relationship between memory stocks, tracked through a DRAM ETF, and the Mag 7 hyperscalers flipped negative as soon as Google announced its $80 billion equity issuance. That is not a gradual drift. It is a step change, and the external evidence confirms both the trigger and the rotation that followed.
Google’s equity offering closed on June 3, 2026, a transaction confirmed by CNBC reporting on the deal’s Goldman Sachs-led structure. At $80 billion, the issuance is large enough to register as a macro event in its own right, not merely a corporate financing decision. When a company of that scale sells equity at that size, it does not just affect its own stock. It reprices expectations for the entire cohort it is associated with and, as it turns out, for the adjacent assets that had been moving in parallel with that cohort.
The mechanics of the flip are not mysterious once the numbers are in view. Benzinga, BigGo Finance, Business Insider, and Defiant Capital Group all documented what happened in the days following the announcement: outflows from the MAGS ETF exceeded one billion dollars, while memory and chip stocks moved in the opposite direction. The Mag 7 group as a whole gained only around 1.1 percent in the period, while memory stocks surged. Two assets that had been rising together started moving apart.
And then basically as soon as Google announced the offering of the $80 billion equity issuance, that positive correlation has flipped negative. Jack Farley
What Farley identified is the non-obvious part of this story. The surface reading of a large equity issuance is dilution, a headwind for the issuing stock. The less obvious consequence is what it signals about the capital structure of the AI buildout. When a hyperscaler of Google’s size goes to the equity market for $80 billion, it is announcing, in the clearest possible terms, that the capital requirements ahead are larger than internal cash flow alone can cover. That is a different signal for memory suppliers than it is for the issuing company’s shareholders. For DRAM, it is not dilution. It is demand confirmation at scale.
The correlation between hyperscaler equities and memory stocks had made intuitive sense when both were rising on the same underlying thesis: AI spending would lift all hardware. That thesis is not wrong, but the equity issuance introduced a wedge. Shareholders in Mag 7 names now face dilution and the possibility that returns will be shared with new capital. Memory suppliers face the opposite: more confirmed demand, no dilution, and a buyer base that has just signaled it intends to keep building.
That wedge is what a correlation flip looks like at the asset level. The two trades had been bundled together in practice, treated by many investors as expressions of the same bet. The Google issuance forced a distinction that the prior price action had papered over. Investors who had held both as a single AI infrastructure position were suddenly holding two positions with different risk profiles and different directional implications from the same catalyst.
Whether the inversion persists depends on how the rest of the Mag 7 finances its coming capital expenditure cycle. If Google’s move presages similar issuances from other hyperscalers, the wedge between hyperscaler equity and memory hardware could widen further. If the issuance was idiosyncratic, the correlation may reassert. But the event itself has already demonstrated that the bundled trade was always more fragile than steady co-movement suggested. Farley’s observation that the flip was near-instantaneous is the detail that matters most. Markets did not wait for earnings revisions or analyst notes. The repricing happened at announcement, which means participants understood the structural implication immediately. The question now is how durable that understanding turns out to be.