SNOW is showing positive signals from all four of its models, an impressive feat. The current long/flat decision is actually driven solely by the ML model, which has proven to be the most robust model for SNOW given its history. (Note that it may not be the case in the future.)

The ML model correctly dodged the dramatic SaaSapocalypse from late November and re-entered the long position in late May. It would have saved investors a lot of money over the past six months.

Based on the simulated history, the Neural model appears more preemptive than the ML model. Therefore, Neural could flash the first sign of the name topping before the ML turned flat. The trend is very favourable for SNOW, so I’m surprised if it turns flat in the near term.

Snowflake sells a managed data platform — storage, a query engine, data sharing and, latterly, AI/ML tooling — consumed on a credit basis rather than a per-seat licence.

The unit of sale is a compute credit. The customer runs a workload, Snowflake bills consumption, and Snowflake, in turn, pays Amazon Web Services, Microsoft Azure or Google Cloud for the underlying infrastructure. This economic model caps Snowflake’s gross margin below 70% because it has to pay infrastructure rent to these hyperscalers.

Therefore, revenue is overwhelmingly consumption-based. Customers typically sign multi-year capacity commitments and then draw down credits as workloads run. This means recognised revenue tracks actual usage rather than contracted seats.

As Snowflake and the hyperscalers improve price/performance, the cost per query falls, but the company must sell more compute units simply to hold revenue flat.

Growth comes from three volume sources:

However, it should be noted that AI products carry a lower gross margin than the core platform, as Snowflake also pays higher rent to its infrastructure suppliers.

We will do a refresh if SNOW remains in our portfolio after the next earnings.

Snowflake’s revenue in Q1 reached $1.4 billion ($1.3 billion from product lines) in Q1, up 34% YoY. This is the fastest growth since the end of 2023.

Remaining performance obligations (RPO) grew 38% YoY against 34% in Q1 last year. Net revenue retention rate rose to 126%, meaning that existing customers are spending more with Snowflake.

779 customers now spend more than $1 million on a trailing 12-month basis, with 46 crossing that threshold in Q1 versus 26 a year ago; 64 customers now exceed $10 million, with 8 added in the quarter.

Management framed the quarter as the moment AI stopped being a tailwind narrative and became a measurable revenue engine.

There is a three-part flywheel: AI pulls workloads onto the core platform, Cortex Code and Snowflake Intelligence monetise directly, and their use drives more core consumption.

One thing that makes me slightly cautious about the AI narrative is that management provided no explicit AI revenue, AI as a percentage of product revenue, or per-account spend uplift during the earnings call, even when asked directly.

Guidance:

After making 67% gross profit, Snowflake’s net margins are eaten by high R&D costs and SG&A costs, creating a negative operating margin. The stock-based compensation is a hefty $403 million. We get some free cash flow back as a result.

The SBC dollar figure is pretty high - $402 million vs 1.4 billion revenue, but it has not grown substantially vs the previous year.

Snowflake enjoys a cash balance of $2.1 billion (current debt negligible), which allows it to buy back shares regularly.

Here are the key financial metrics from the past four quarters. I designed the table this way so we can focus solely on the key top-line, profit, and cash flow metrics.

Here are Wall Street's forward estimates. The consensus is that Snowflake will grow at a slower pace in the coming quarters, but still in the high 20s.

Snowflake rents out a data warehouse, a place where a company keeps its business data and pays only for the computing it actually uses, like a household paying for electricity by the unit rather than a flat rent.

Base case

The first AI consumption wave matures, and revenue lands in line with the platform estimate of $7.65bn for the year to January 2028. Growth of about 26% YoY is still excellent, but the multiple settles back to 14x forward sales. Revenue compounds into a shrinking multiple and the two cancel out.

The trigger is an in-line September quarter, with product revenue inside the $1,415m to $1,420m guide and net revenue retention holding near 126%.

Bull case

Agentic AI usage proves durable rather than experimental. Cortex-driven consumption carries revenue to about $8.1bn for the year to January 2028 and the market keeps paying 17.5x forward sales. That will put the stock price appreciation on par with revenue growth.

Bear case

One soft consumption quarter rerates the stock lower. Because customers can throttle usage within days rather than waiting for a renewal, growth decelerates toward 20% YoY, revenue for the year to January 2028 comes in near $7.2bn, and the multiple compresses to 10x forward sales.

Snowflake currently trades near its 3-year high in terms of valuation, but this is justified by the fastest pace of growth in the recent quarter.

Snowflake is likely to sustain its 30% growth rate, given the increasingly intensive use of AI workloads by its customers and management’s high credibility in execution. That makes 17x NTM EV/Sales pretty fair, a multiple seen in early 2024 and not out of whack with the historical valuation range.

That makes the current price fair.

Our DCF model judges the stock to be priced very optimistically. Today’s price, given the high risk-free rate, needs to see SNOW effectively stay on its 20%-30% growth for the majority of the next decade while simultaneously raising its profit margins significantly (GAAP EBIT margin to high 30%s).

This is possible, but very, very tight. While management could certainly outperform on revenue growth (even keeping the 30% growth longer than expected), profitability would be a major challenge.

Investors today are paying for a high-growth, accelerating, and capital-light business with an excellent management team that is executing very well. However, they are also buying a business model that is less predictable than the seat-based model, both in terms of revenue and cost.

The stock price accounts for further revenue acceleration with improved profitability over time. But there are key risks that Snowflake’s suppliers, the hyperscalers, compete directly with Snowflake on the data storage front. Snowflake’s “high switching cost” defensibility does not necessarily last over time, as challenger businesses can help customers migrate their data layer with increasing ease.

The 2026 rally, R2, has already surpassed the entire rally from 2024H2 to 2025H2 in just four months. The uptrend remains intact, with the price well above all key moving averages, meaning it’s not a good time to short the stock. However, SNOW looks more like a “ride-the-winner” trade than an “initiate now” trade. Therefore, if the tide turns and our Systematic Portfolio drops the name, I would not hesitate to follow.

Overall, Snowflake has seen impressive performance in the past few months, amidst revenue acceleration and multiple expansion. Its valuation is high vs history but still defensible. It is not a name to chase, but one that we can ride with the portfolio.