Charts of the Week: Retail to the Moon
Borrowing Bigly for AI; Selling Software to Uncle Sam; Grocery is an advertising business
Is this your brand on Milled? Claim it.
Charts of the Week: Retail to the MoonBorrowing Bigly for AI; Selling Software to Uncle Sam; Grocery is an advertising business
America | Tech | Opinion | Culture | Charts
Retail to the moonWhile many (if not most) of the big pandemic themes have reverted to their pre-pandemic trend, one of them most certainly has not: retail participation in the stock market took off during the lockdown, and has shown no signs of slowing. If anything, retail is playing a bigger role in moving markets than ever before. And retail is buying with more leverage, more frequency, and generally speaking, more optimism than ever before, as well. Citadel Securities, which sees roughly 40% of US retail order flow, reports that May is on pace to be the most active month for retail cash equity volume on record. Retail cash is tracking ~12% above the January 2021 meme-stock peak, and perhaps more striking, retail option volumes are on a similar record-breaking trajectory, with average daily contracts pacing roughly 60% above the historical monthly average.
Unsurprisingly, retail option-buyers are going to where the action is (or maybe the action is going where retail is), with semiconductors becoming an especially popular target:
Retail option volume in semiconductors is now running at roughly 2.8x the post-2020 monthly average, and about a third higher than just a month ago. Until very recently, trading options was generally considered the exclusive domain of trained professionals. There’s a lot of math—market makers like Citadel are chock full of some of the mathiest people in the world—and there’s a lot of risk, as well. Sure, it’s a relatively cheap way of getting exposure to changes in stock prices . . . but if those prices move against you, then instead of owning a slightly depreciated share of equity, you own, well, nothing. Retail getting risky, however, is very much the broader theme. Take levered ETFs, as another example.
According to Goldman, retail’s share of volume on index ETFs gets bigger the more leverage those ETFs take-on. Retail accounts for ~20% of volume on triple-levered S&P ETFs, and a full 25% of triple-levered Nasdaq ETFs. Leverage, of course, amplifies your returns on the way up . . . and does the same to your losses, on the way down. Margin tells a similar story, with margin balances in retail trading accounts hitting all-time highs:
Combined margin balances at Interactive Brokers, Robinhood, and Schwab have gone vertical, now at more than $250 billion—which is 5x(!) more than it was before the pandemic. As a % of customer assets, margin is still only ~1.7%, but that too is at or near the all time high, with margin running at roughly 1.7% of customer assets (the highest reading since 2021). As an interesting aside, not all retail account holders are created equal:
Interactive Brokers accounts trade far more than other retail accounts, with ~4x the daily trading volume. In all events, good or bad, there’s no question that retail investors–or perhaps a small subset of them–are playing an historically large role in markets. Retail is not only more active, but it’s taking on more risk, and recently that risk almost uniformly skews to the upside:
While Retail slowed its roll a bit during the thick of the Hormuz crisis, retail trading came roaring back to life in mid-April and appears to have brought the index with it. From April through most of May—seven full weeks—there was not a single day where retail order flow favored “sell.”
According to Scott Rubner of Citadel, retail-the-perma-bull represents something of a shift. To paraphrase, through most of this year, retail was a contrarian buyer (selling rallies and buying dips). That made retail something of a stabilizer through April and May. But in the last few weeks, that behavior flipped. Retail is now buying on both up days and down days, and they’re using historic levels of leverage to drive the rally up-and-to-the-right. Of course, as Rubner also points out, all that one-sided push creates “flow fragility,” i.e. the risk of massive one-sided unwinding in the other direction. Stay safe out there. Borrowing Bigly for AIIn his recent podcast with David Haber, Marc Rowan, the CEO of the PE/PC giant, Apollo, said “every dollar since the invention of fire is going into this AI capex build-out – and equity can’t fund it all.” Rowan’s point is that debt is increasingly a major part of the AI infra story, and he’s not wrong. If, in the early innings, the hyperscalers funded capex with profits, they have now shifted an increasing share of the load to debt.
Hyperscaler bond issuance has increased dramatically over the past two years, with ~$150B already printed in 2026. That’s ~35% more debt than last year, and there is still half the year left to go. AI-related debt isn’t just a big deal for the hyperscalers, it’s a big deal for the credit markets, writ large:
In 2026, roughly half of investment grade net issuance and ~40% of high yield bonds are tied to AI capex. AI continues to be the everything-story for everyone. “IT Capex” is approaching 40% of all Capex for the entire S&P:
It’s unclear what exactly capital markets would be doing now without AI. Tending to their knitting, perhaps. Interestingly, despite all this borrowing, the risk outlook for hyperscalers hasn’t changed all that much. Hyperscaler IG credit has historically been perceived as much safer than the general universe of IG credit (reflected in their lower spreads), and while the gap has definitely narrowed, they’re still considered “safer,” even if not “much safer”:
If you exclude Oracle (by far the least well-capitalized of the hyperscalers), hyperscaler IG credit is still less expensive than IG credit more broadly. Including Oracle, that’s no longer the case, but again, Oracle is an outlier in how much it’s borrowed relative to its overall cashflows. The point being that however one feels about putting an increasing share of capex on credit, if there are any firms that can stand to add a little leverage, it’s the hyperscalers, who are still much less leveraged than most:
Hyperscaler debt/equity ratio only recently caught up with the rest of Nasdaq 100, and still trails the rest of the S&P by a good deal (albeit about half as much as before). That isn’t to say that all this borrowing is the proverbial “nothingburger,”--far from it, as per above, it’s basically the prime mover in credit markets right now–but, there’s a difference between a big-time borrowing binge, and a big-time borrowing binge off a very low base. Selling Software to Uncle SamAbout a year ago, the US General Service Administration announced FedRAMP 20x, an initiative “focus[ed] on innovating alternative approaches to make automated authorization [for tech providers] simpler, easier, and cheaper while continuously improving security.” FedRAMP refers to the Federal Risk and Authorization Management Program, which is generally responsible for providing the security authorization required to sell software to the government. In the past, FedRAMP authorization involved a cumbersome, time-consuming process, but FedRAMP 20x has changed the game:
The agency’s first 300 approvals took a little over a decade, but under FedRAMP 20x, it approved ~140 new products in 2025 alone. The program overhaul comes not coincidentally with the government’s efforts to invest in AI—only a few months after FedRAMP 20x was announced, the government announced ~$800M in new contracts with the leading frontier model companies, as well as big contracts with Palantir and Anduril. But non-AI companies are benefiting, as well (and, in fact, comprise ~80% of the approvals in 2025). Look, it’s definitely still the case that selling software to the government is no easy task. Procurement cycles are long, and the security requirements are more onerous than the private sector. But, the good news is that at least one hurdle has become substantially lower than before. Grocery is an advertising businessOne thing I learned from Byrne Hobart a few years ago is that Grocery is really an advertising business. Not totally, of course, but certainly in a lot of ways. Even before the advent of ecomm, groceries were busily monetizing eyeballs with their own form of ranked-search. All those tasty, low-margin goodies, like rotisserie chicken and extra long hot dogs, serve a singular purpose: get shoppers in the store, so that grocers can make the big bucks by selling premium shelf-space to CPG companies. Oh, you want your high-margin namebrand jarred salsa and tortilla chips to sit mere feet away from the checkout line, with their own special display? We can handle that, but it’ll cost you. With the advent of ecomm, the grocery-cum-advertising industry, has taken things to a whole new level. Loaded to the gills with first-party consumer data, and a steady flow of high-intention shoppers, big name grocers are selling (very high margin) product-placement like never before. Take a look at Walmart:
Walmart’s US advertising revenue growth has been accelerating (more or less) since Q4 ‘24. It’s now growing ~45% in the latest read. Not bad for what’s already the second-largest retailer in the world. Walmart’s other big recent success has been its Costco-like membership program:
Membership income growth had been slowing down to a still impressive ~15% yoy, but it perked up slightly last quarter. These aren’t trivial line-items, either. Membership and advertising income now account for ~1/3rd of Walmart’s operating income (despite being a tiny fraction of overall revenue). Walmart has certainly benefited from other cyclical factors, as well. Rising prices, and slowing income growth have led to more higher-income shoppers “trading down,” such that Walmart has been expanding its customer-base upmarket since at least 2023. But, there’s a reason why these major consumer brands want you to use their apps–and not just Walmart—add Kroger, Instacart, and Target, just to name a few. Groceries have always been in the advertising business, and they know an opportunity (and first party data) when they see one. As an interesting aside on the subject of affordable groceries, according to Consumer Reports, Costco remains the king of low prices:
Nationally, Costco runs ~21.4% cheaper on average than Walmart. Costco famously has a different sort of grocery business model. It sells its wares with the lowest markup possible, and makes the majority of its profit on a tiny slice (~2%) of its overall revenue: membership fees. Of course, those too come with a trove of first-party data to inform other sorts of business decisions, and not necessarily advertising . . . but even Costco has started to dip its toes into the advertising business. Back in March, Costco’s CFO offered some color:
The CFO also clarified that for Costco, it’s of course not about the revenue so much as the customer experience. Sure, why not? As long as the hot dogs stay cheap.
This newsletter is provided for informational purposes only, and should not be relied upon as legal, business, investment, or tax advice. Furthermore, this content is not investment advice, nor is it intended for use by any investors or prospective investors in any a16z funds. This newsletter may link to other websites or contain other information obtained from third-party sources - a16z has not independently verified nor makes any representations about the current or enduring accuracy of such information. If this content includes third-party advertisements, a16z has not reviewed such advertisements and does not endorse any advertising content or related companies contained therein. Any investments or portfolio companies mentioned, referred to, or described are not representative of all investments in vehicles managed by a16z; visit https://a16z.com/investment-list/ for a full list of investments. Other important information can be found at a16z.com/disclosures. You’re receiving this newsletter since you opted in earlier; if you would like to opt out of future newsletters you may unsubscribe immediately.
|
