Snowflake's $3.5 Billion Convertible Rattles Cloud Stocks
A proposed $3.5 billion convertible note from Snowflake knocked cloud software names lower on Monday, but the dilution math depends on structure — and the intraday tape already told a different story.
Snowflake fell about 4% on Sept. 28, 2026 after proposing a $3.5 billion convertible note offering, dragging Oracle down 3% and Datadog down 4% intraday before partial recoveries, with SNOW last quoted at 330.93, down 1.49%.
Snowflake (SNOW) proposed a $3.5 billion convertible note offering on Monday, and the software complex took it personally. The stock slid about 4% on the news, Oracle (ORCL) dropped 3% and Datadog (DDOG) fell 4% — a sympathy move that had less to do with any of those companies' operations than with what a jumbo convertible says about how expensive equity-linked funding has become for the cloud cohort.
By the last trade at 16:47 GMT on Sept. 28, 2026, the damage had been partly repaired. Snowflake was quoted at 330.93, down 1.49% from a prior close of 335.94, after touching an intraday low of 316.63. Oracle changed hands at 133.66, off 2.51% against a 137.10 close, having bottomed at 131.58. Datadog had done something different entirely: it was up 1.11% at 271.10, after trading as low as 255.50 — a swing of roughly 6.1% off the floor, and a sign that the initial markdown was a reflex rather than a verdict.
What a convertible note actually does to shareholders
A convertible note is a bond that the holder can exchange for stock at a set price. The issuer gets cheaper money than a straight bond would cost, because the buyer is being paid partly in optionality. The trade-off is potential dilution: if the shares rise above the conversion price, those bonds become equity, and every existing shareholder's slice gets smaller.
That is the reflexive fear that shows up in the tape within seconds of a headline like this one. But the reflex frequently overshoots, because the structure of modern convertible deals is built specifically to blunt the dilution. Two features do the work:
- Conversion premium. The conversion price is typically set well above the stock price on the day of pricing. The bonds only become shares if the equity climbs materially from here — which is to say, dilution arrives only in the scenario where shareholders have already been paid.
- Capped calls. Issuers routinely buy call options on their own stock alongside the deal, using a slice of the proceeds. Those options offset the shares that would be delivered on conversion, effectively raising the price at which real dilution starts. The cost is that some of the $3.5 billion never reaches the balance sheet.
There is a second-order effect that traders watch closely on pricing day. The hedge funds that buy convertibles are usually not making a directional bet on the company; they are buying the embedded option and shorting the stock against it to isolate the volatility. That delta hedging creates mechanical selling pressure into the deal — pressure that has nothing to do with anyone's view of the business and that tends to unwind once the book is closed. That dynamic is a reasonable candidate for the gap between Snowflake's 316.63 low and where it actually sits now.
Why Oracle and Datadog got dragged in
Neither Oracle nor Datadog issued anything on Monday. They were repriced because investors treat the cloud and data-infrastructure names as a single funding bloc. When the fastest-growing member of that bloc reaches for $3.5 billion in equity-linked paper, the market reads it two ways at once: as evidence that capital needs are scaling with AI-era compute commitments, and as a signal about what the equity-linked market will bear before the terms get punitive.
Oracle is the more logical sympathy trade of the two. It is the name in this group most associated with heavy, debt-funded capacity buildout, and it closed the session's move at 133.66 — down 2.51%, and the weakest of the three on the day. Its bounce off 131.58 was the shallowest as well, around 1.6% off the low, which suggests the selling there was more considered than the knee-jerk that hit Datadog.
6% off the low, which suggests the selling there was more considered than the knee-jerk that hit Datadog.
Datadog is the counter-example. It fell 4% on a headline about another company's capital structure and finished the stretch up on the day at 271.10, within a whisker of its 271.28 session high. That round trip is the clearest evidence available that Monday's cloud selloff was positioning, not fundamentals.
The broader tape was already soft
Context matters here, because none of this happened in a vacuum. The S&P 500 tracker (SPY) was at $767.04, down 0.56%, and the Nasdaq 100 (QQQ) at $737.81, off 0.90% — growth-heavy indexes leading the downside, as they do on days when long-duration cash flows get discounted a little harder. The Dow 30 (DIA) held up better at $515.12, down 0.46%. On a session where the Nasdaq was already the weakest of the three benchmarks, a large convertible out of a high-multiple software name was always going to land harder than it would on a risk-on day.
As 24/7 Wall St noted in flagging the move, the deal's unusual structure is the part worth interrogating before anyone concludes the worst about dilution.
What to check when the terms land
Proposed offerings are not priced offerings. The $3.5 billion is a target; the coupon, maturity and conversion premium get set after the book builds, usually within a day. Several specifics will determine whether Monday's markdown was justified:
- The conversion premium. A wide premium means the shares only convert after a substantial rally. A narrow one means dilution is a live near-term possibility.
- Whether capped calls are included, and how high the cap sits. This is the single biggest determinant of effective dilution, and it is disclosed in the pricing release.
- Use of proceeds. Refinancing an existing convertible is a materially different story from funding new compute capacity or buying back stock. If a portion goes to repurchasing shares concurrently with the offering — a common pairing — the net share count impact shrinks further.
- Upsizing. Strong demand often pushes the final size above the proposed figure, along with a greenshoe option for the underwriters. That would signal the market is happy to fund Snowflake at attractive terms, even if the headline number looks larger.
The through-line for the sector is simpler than the securities math. Cloud and data platforms are committing to capital spending on a scale that operating cash flow alone will not cover, and the convertible market — cheap, fast, and forgiving on covenants — is where that gap gets bridged. Investors should expect more of these, not fewer. The useful question on each one is not whether dilution exists, but where the conversion price sits relative to the share price, and who is really on the other side of the trade.
Key facts
- SNOW last trade: 330.93, -1.49%, as of 16:47 GMT Sept. 28, 2026
- Offering size: $3.5 billion proposed convertible notes
- Sympathy moves: Oracle 133.66 (-2.51%); Datadog 271.10 (+1.11%)
- Benchmarks: QQQ $737.81 (-0.90%); SPY $767.04 (-0.56%)
Frequently asked questions
How much is Snowflake raising?
Snowflake has proposed a convertible note offering of $3.5 billion. The figure is a target rather than a completed deal — final size, coupon, maturity and conversion terms are set once the order book builds, and strong demand can push the final amount above the proposed number through an upsize or an underwriters' option.
Why does a convertible note cause the stock to fall?
Two reasons. First, convertible bonds can turn into shares, which dilutes existing holders if the stock rises past the conversion price. Second, the hedge funds that buy convertibles typically short the issuer's stock to isolate the embedded option, creating mechanical selling pressure into the pricing that often unwinds once the deal is placed.
What is a capped call and why does it matter here?
A capped call is an option the issuer buys on its own stock, funded from the offering proceeds, to offset shares that would be delivered on conversion. It effectively raises the price at which genuine dilution begins. Whether Snowflake includes one, and where the cap is set, is the single biggest variable in the dilution question.
Where did Snowflake, Oracle and Datadog trade on Sept. 28?
At the last trade at 16:47 GMT, Snowflake was 330.93, down 1.49% from a 335.94 prior close, after a session low of 316.63. Oracle was 133.66, down 2.51% from 137.10. Datadog was 271.10, up 1.11% from 268.13, having recovered from an intraday low of 255.50.
Why did Oracle and Datadog fall on Snowflake's news?
Neither company issued anything. Investors treat cloud and data-infrastructure names as a single funding bloc, so a large equity-linked raise by one member gets read as a signal about capital needs and financing costs across the group. Datadog's full recovery to positive on the day suggests that read was a reflex rather than a fundamental judgment.
What should investors watch next on this deal?
The pricing release. Key items are the conversion premium relative to the current share price, whether capped calls are attached and how high the cap sits, the stated use of proceeds — refinancing versus new capacity versus a concurrent buyback — and whether the final size is upsized above $3.5 billion.
