Retail bought $27bn of Nvidia in a year. That is inventory, not a floor

Over the trailing year, individual investors made a net $27 billion of purchases in Nvidia, more than in any other Magnificent 7 stock, according to Vanda Research flow data reported by Cryptobriefing ^[1][2]. At various points Nvidia's weight in a typical retail portfolio exceeded 10% ^[3], which means the marginal bid in the market's largest AI position is household money concentrated in a single ticker.

The pace matters more than the headline number. Spread evenly, $27 billion over twelve months is about $2.25 billion a month ^[10], roughly $520 million a week ^[11]. Vanda, which specialises in tracking US retail equity flows and is widely used by institutions to gauge positioning ^[8], has documented single-day net inflows in the hundreds of millions and accumulation that stacks up quarter after quarter, and has repeatedly labelled Nvidia the "retail king" ^[4][5]. The firm has also flagged periodic rotations of retail capital into other names, Tesla most notably, but describes them as temporary ^[6].

Sustained one-way flow of that kind usually gets described as support. Inventory is the better word. A position above 10% of a portfolio is not a diversified holding; it is the portfolio's dominant risk, and it sits with owners who have no mandate, no rebalancing discipline imposed on them, and sell triggers that have nothing to do with data centre orders: a job loss, a tax bill, a margin call, or simply the moment the loss stops feeling temporary. Institutions that hold the same stock are at least required to trim into strength. Households are not, which is why their concentration builds quietly and unwinds all at once.

The clearest evidence of how this data is actually used comes from who buys it. Institutional investors monitor retail flow feeds such as VandaTrack to understand positioning and to anticipate volatility in high-profile stocks ^[7]. That is not a search for a floor. It is a search for the crowded side of the boat, which is useful information precisely because the crowd moves together.

Two caveats on the material. The supplied reporting is a secondary account: Cryptobriefing publishing via designrush.com, citing Vanda rather than a primary Vanda note ^[9]. And it gives no dollar figures for retail buying in the other six Magnificent 7 names, no start and end date for the trailing-year window, and no comparison against Nvidia's total traded volume ^[12]. So $27 billion is a headline without a denominator. Against the daily turnover of the most heavily traded large cap in the world, retail is not setting the price on any given day; against the incremental flow that has to keep arriving to hold a valuation up, it is not a rounding error either.

What to watch: whether Vanda's weekly prints show retail net selling on down days rather than the dip-buying reflex that has held so far, since a first genuine capitulation week would be the real test of the bid. Watch whether the 10%-plus portfolio weight comes down through rotation into other names or only through price. Watch whether the Tesla rotations start lasting longer than the source describes as temporary ^[6]. And watch for a dated, primary Vanda publication confirming the $27 billion window before anyone builds a thesis on it ^[1].