Snowflake directors sell EUR 33.5m while cloud peers keep climbing

Snowflake Snowflake Inc. is not being sold into a vacuum. The cloud infrastructure complex is still getting paid for AI-related demand, and the biggest platforms are still posting the kind of growth that keeps the whole category in the market’s good graces. Google Cloud revenue rose 82% year over year to $24.8 billion in the most recent reported quarter, AWS grew 37% to $42.2 billion, and Microsoft Azure increased 43%. Those are not sleepy numbers. They are the kind that keep capital flowing into adjacent data and analytics names, even when the tape gets choppy.

That is the useful frame for Snowflake. The company sits in the data platform layer of that stack, where consumption-based usage tied to AI workloads has supported the bull case. The stock has also been volatile enough to make any insider sale look larger than it would in a dead market. Snowflake closed at $331.48 on September 9, after trading between roughly $305 and $385 over the prior week. So the directors did not sell into a flat line. They sold into a stock that had already been moving around with some force.

The two sales, and why the size matters more than the headline

The filing set is straightforward. Director Mark Garrett sold 50,000 shares at $350 each on September 4, for a euro-normalised filing value of EUR 15,058,749.38. Director Michael L. Speiser sold 50,741 shares at approximately $353.87 each, for about EUR 17.96m, and the transactions were executed under a pre-adopted Rule 10b5-1 trading plan. Together, the two sales total roughly EUR 33.5m.

The market value context matters here. InsiderTrades data pegs the combined filing value at about 0.02% of Snowflake’s market value, which is small in percentage terms and large in absolute terms. That is the kind of sale that does not change the capital structure, but it does tell you the board is willing to take chips off the table at a time when the stock has already had a strong run. The company’s market cap in the dossier sits near EUR 97.97bn, so this is not a distressed exit. It is a deliberate reduction in exposure.

The fact that Speiser’s trades sat inside a 10b5-1 plan matters too, but only so much. A plan removes some of the discretion from the timing. It does not remove the economic fact of the sale. Garrett’s filing is the cleaner read because it is an open-market disposition by an operating director, and our scoring leans on that sort of filing when it is paired with a cluster. Here, the cluster is not a single dramatic print. It is a set of recent declarations that shows multiple insiders trading the name within a month.

Snowflake against Datadog and MongoDB

The comparison that makes the most sense is Datadog. Both names sit in adjacent infrastructure software territory, both are exposed to enterprise data demand, and both have been treated as premium software franchises when the market wants growth with a cloud wrapper. Datadog shares have traded near levels implying a market capitalization around $80bn, while Snowflake has sat closer to $115bn. That gap is not trivial. It tells you Snowflake still carries the larger platform premium, and it also tells you the market has been willing to assign a heavier multiple to the data layer than to observability when the narrative is working.

MongoDB belongs in the same conversation for a different reason. It is another infrastructure software name where the market has had to decide whether usage growth and platform relevance justify the valuation. Recent insider sales at MongoDB and Datadog have shown that Snowflake is not the only name where directors and officers have been trimming. That matters because it keeps this filing from looking like a one-off alarm bell. It looks more like a sector habit. When software names are priced for durable AI and cloud demand, insiders often choose that moment to sell into strength.

Still, Snowflake is the richer stock in the comparison set, and that changes the burden of proof. A company with a larger market cap and a more visible AI-linked story can absorb insider selling more easily than a smaller, less liquid peer. But it also has farther to fall if the market decides the growth path is not as clean as the multiple implies. That is the tension here. The directors are selling a name that still commands a premium, and the premium is doing some of the work that the business itself has to keep earning.

The cluster, and what it does not tell you

InsiderTrades data marks this as a cluster, with three distinct insiders and 12 recent declarations. That is the part that deserves attention, because a lone sale can be mechanical and a cluster can reflect a broader willingness to reduce exposure. Here, the recent declarations include multiple September filings, and the pattern is not hidden. Garrett sold, Speiser sold, and the company has had additional sales by other executives earlier in September according to the grounded research. You do not need to overread that. You do need to notice it.

The historical cohort read is mixed, which is exactly how it should be for a mature mega-cap bucket. For director-level buys at mega-cap names, the 90-day win rate is 47.2% and the average 90-day return is 0.58%, with a 365-day average return of 87.53% across a sample of 5,264. That is historical cohort data, not a promise about Snowflake and not a forecast for this trade. It does, however, keep you honest. Director activity at this size bucket has not been a clean short-term edge. Sometimes it works. Sometimes it does not. The point is that the filing belongs in a pattern, not in a prophecy.

The internal score rationale also fits the same frame. The filing was made by an operating director, it came as part of an insider cluster, and it was sized at about 0.02% of company value. Those are the ingredients our scoring leans on. They do not make the sale bearish by themselves. They make it worth reading in context, especially when the stock has already had a wide trading range and the sector is still rich enough to invite profit-taking.

Macro noise, rich software, and why timing matters

The broader market backdrop is not helping high-multiple software names much. A stronger-than-expected August U.S. jobs report showed 162,000 additions against lower expectations, with unemployment at 4.1%. Treasury yields were elevated near 4.78% to 4.84% on the 10-year note, and oil prices were rising on geopolitical tension. That mix kept pressure on major indexes and made the market less forgiving of stretched valuations. Snowflake does not trade like a bond proxy, but it does trade like a name whose multiple is sensitive to discount rates and risk appetite.

That is where the comparison with Datadog matters again. In a softer macro tape, premium software names can still work if the growth story is strong enough. But the market tends to become more selective. It rewards the cleanest execution and punishes anything that looks like narrative fatigue. Snowflake has the advantage of being tied to cloud data and AI demand, which is still one of the market’s favorite themes. It also has the disadvantage of being priced like a company that has to keep delivering. The directors’ sales do not change that. They simply arrive at a moment when the market is already asking harder questions about duration and valuation.

You can see the same dynamic in the peer group. Datadog and MongoDB have both had insider selling reported in the same period, which suggests the behavior is not isolated to Snowflake. That does not make the sales harmless. It makes them legible. When a cluster of software insiders chooses to sell while the sector is still being rewarded for AI exposure, the message is usually about valuation discipline rather than a sudden collapse in business quality. Usually. Not always.

Snowflake’s fundamental score versus the market’s patience

InsiderTrades data gives Snowflake a fundamental score of 34, with a rank of 21,906 out of 29,063. The component values in the dossier show value at 37 and quality at 32, with growth not provided. That is not a flattering profile, and it should not be dressed up as one. Snowflake is still a premium platform story, but the internal fundamental screen is not screaming strength. That matters because insider selling is easier to dismiss when the business is accelerating cleanly and the valuation is still modest. Snowflake is not in that camp.

The market is still willing to pay for the name because the cloud and AI backdrop remains supportive. But patience is not infinite. If the stock is already trading in a wide band, and if the directors are selling into that band, then the burden shifts back to execution. The next quarter will matter more than the last filing. So will any evidence that consumption growth is broadening beyond the most obvious AI use cases. Snowflake has to keep proving that the platform can monetize the current wave without relying on a single narrative to do all the work.

The comparison with Datadog is useful here because it shows how the market can keep two adjacent software names in the same premium bucket while still treating them differently. Datadog’s market cap near $80bn is smaller, but the market has also been willing to pay for its observability footprint. Snowflake’s larger cap near $115bn gives it more room, but also more expectations. Directors selling EUR 33.5m of stock does not break the story. It does remind you that the people inside the boardroom are not treating the current price as too cheap to touch.

What to watch after the September filings

The next thing to watch is whether the selling stays clustered or fades back into routine 10b5-1 activity. A single plan-driven sale is one thing. Multiple September dispositions by different insiders is another. If the pattern extends, the market will start to read it as a more deliberate reduction in exposure rather than a calendar event. If it stops here, the filing will look more like profit-taking after a strong move in a volatile stock.

The stock itself is the other tell. Snowflake closed at $331.48 on September 9 after a week that ranged from roughly $305 to $385. That kind of range tells you the market is still willing to reprice the name quickly. If the shares hold up despite the insider sales, the market is saying the cloud and AI story still dominates. If they fade back toward the lower end of that band, the sales will look better timed and the valuation debate gets louder.

For now, the comparison with Datadog is the cleanest way to read Snowflake. Both names sit in the same premium software neighborhood. Both have seen insider selling. Both are tied to enterprise infrastructure demand that the market still likes. Snowflake just carries a larger valuation burden, and that makes every director sale a little more visible. The filing is not a verdict. It is a reminder that even in a strong cloud tape, some insiders prefer to sell strength rather than wait for the market to do the work for them.

This is not investment advice.