Snowflake told Wall Street two things on the same morning: quarterly revenue had grown 33% to $1.39 billion, and the company was handing Amazon Web Services $6 billion over the next five years. The announcement landed on May 27, 2026, and the stock jumped as much as 36% that day, according to CNBC’s coverage of the move, one of the sharpest single-session swings a major cloud software vendor has posted this year.
The timing wasn’t an accident. Pairing an earnings beat with the largest AWS commitment in company history let Snowflake tell one story instead of two: enterprises are buying AI infrastructure faster than most models predicted a year ago, and Snowflake wants to be the software layer sitting on top of it. A little over a week later, Pinterest signed its own $4 billion AWS agreement. About five weeks before that, Anthropic locked in a 10-year, nine-figure-billion compute commitment with the same cloud provider. By late June 2026, a pattern that looked like a one-off in early spring reads like a trend. AWS is closing some of the largest infrastructure contracts in its 20-year history, one AI-hungry customer at a time.
Here’s what the Snowflake AWS deal actually includes, why it arrived bundled with an earnings report, and what it signals about where enterprise cloud spending is heading for the rest of 2026.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What the Snowflake-AWS Agreement Actually Covers
Snowflake and AWS call it a Strategic Collaboration Agreement, the same contract structure AWS has used for years to lock in joint spending, marketing, and engineering commitments with major software partners. The version Snowflake announced on May 27, 2026 runs five years and totals $6 billion, which the company describes as its largest AWS commitment to date.
The money covers three areas. First, compute: Snowflake is expanding its use of AWS Graviton, Amazon’s custom processor line, to run more of its data warehousing and AI workloads. Second, go-to-market: the deal broadens Snowflake’s presence in AWS Marketplace, the storefront enterprises use to buy third-party software directly against their existing AWS billing agreements. Third, product integration: the two companies are deepening ties between Snowflake’s Cortex AI tools and Amazon’s generative and agentic AI services, so AWS customers can build AI agents that query Snowflake data without leaving the AWS console.
Snowflake’s own announcement says the agreement is meant to “accelerate enterprise agentic AI adoption” and help customers build and deploy AI systems “faster and more securely.” CRN’s reporting framed the agreement as extending Snowflake’s governance reach from stored data to AI actions and agent behavior across a customer’s entire AWS environment, a broader scope than a typical infrastructure renewal.
The Earnings Beat That Set Up the Announcement
Snowflake didn’t arrive with just a headline number. Fiscal first-quarter 2027 results, released the same day as the AWS news, showed total revenue of $1.39 billion, up 33% year over year. Product revenue, the company’s core recurring metric, came in at $1.33 billion, up 34%. Net revenue retention landed at 126%, meaning existing customers spent 26% more, on average, than they did a year earlier.
| Metric | Q1 Fiscal 2027 Value | Year-Over-Year Change |
|---|---|---|
| Total revenue | $1.39 billion | +33% |
| Product revenue | $1.33 billion | +34% |
| Net revenue retention rate | 126% | Existing customers spending 26% more |
| Non-GAAP operating income | $165.8 million | 12% operating margin |
| Stock reaction (per CNBC) | Up to +36% in a single session | Tied to earnings plus AWS news |
Those figures matter because they’re the evidence Snowflake needed to justify a five-year, $6 billion bet. A company growing product revenue 34% a year can plausibly commit half a decade of spend against future consumption. A company growing 10% a year would be making a much riskier wager, and investors would likely have read the same announcement very differently.
Why the Stock Jumped 36% in a Single Session
Moves that large are unusual for a company already worth tens of billions of dollars. CNBC’s reporting attributed the jump to the combination of the earnings beat and the AWS announcement landing on the same day, rather than either one alone. Investors read the pairing as confirmation that Snowflake’s AI products, chiefly its Cortex suite, are converting into real usage rather than staying stuck in pilot-stage testing, which has been the central worry hanging over enterprise AI vendors since the hype cycle began.
The 126% net revenue retention rate tells a similar story from a different angle. It means the average existing Snowflake customer ran more workloads through the platform than a year earlier, not just renewed at flat spend. Wall Street tends to reward that number more heavily than new-logo growth, because retention is harder to manufacture and harder to lose overnight than a splashy signing.
AWS Graviton, Explained
Graviton is AWS’s line of custom-built processors, using ARM architecture instead of the x86 chips from Intel and AMD that have powered most of the world’s servers for decades. Amazon designs Graviton chips in-house specifically for cloud workloads, and AWS has spent close to a decade pushing customers toward them by pricing Graviton instances below comparable x86 options.
For a company like Snowflake, whose entire business model runs on charging customers to store and query data, shaving compute cost per query directly improves gross margin. That’s the mechanical reason Graviton sits at the center of the AWS deal. It isn’t only a marketing commitment. It’s a lever Snowflake can pull to make its own infrastructure cheaper to run, quarter after quarter, without raising prices on customers.
Not a One-Off: AWS’s Billion-Dollar Client List in 2026
Snowflake’s $6 billion agreement is the second-largest publicly disclosed AWS infrastructure commitment of 2026, and it landed in the middle of a run of similar deals from very different types of customers.
| Company | Announced | Disclosed Value | Term | Primary Focus |
|---|---|---|---|---|
| Anthropic | April 20, 2026 | Up to $25B in Amazon investment, $100B+ in Anthropic AWS spend | 10 years | Trainium chips, Project Rainier, 5GW of capacity |
| Snowflake | May 27, 2026 | $6 billion | 5 years | Graviton compute, agentic AI, AWS Marketplace |
| June 4, 2026 | $4 billion | Through 2031 | AI-powered visual search and content discovery | |
| Meta | 2026 (date undisclosed) | Undisclosed | Undisclosed | Tens of millions of Graviton chips for agentic AI |
Anthropic set the tone in April, agreeing to spend more than $100 billion on AWS compute over a decade in exchange for up to $25 billion in new Amazon investment and 5 gigawatts of dedicated capacity, most of it running on Trainium chips through the companies’ joint Project Rainier cluster. Snowflake followed in May. Pinterest signed its own multi-year, $4 billion agreement on June 4, extending an existing relationship to cover AI-powered visual search and content discovery, according to reporting on the Pinterest agreement. AWS CEO Andy Jassy has also said Meta committed to running tens of millions of Graviton chips for CPU-heavy agentic AI workloads, though neither company has disclosed a dollar figure for that arrangement. AWS’s own legal counsel on the Pinterest deal, Sidley Austin, confirmed the firm’s role in structuring that long-term cloud services agreement.
Add it up and AWS locked in well over $130 billion in disclosed multi-year commitments from just four customers within about six weeks this spring. That is not normal deal flow, even for a company AWS’s size, and it says as much about the current state of enterprise AI budgets as it does about any single vendor’s product roadmap. Uber’s earlier Trainium3 chip agreement with AWS shows the same pattern reaching beyond software vendors into companies that simply need cheaper AI inference at scale.
Where AWS Stands in the Cloud Market
| Provider | Est. Global Market Share (Q1 2026) | Position |
|---|---|---|
| AWS | ~29%-30% | #1, still the largest by revenue |
| Microsoft Azure | ~25% | #2, growing fastest on a percentage basis |
| Google Cloud | ~13% | #3, smallest of the top three but accelerating |
| All others combined | ~32%-33% | Fragmented among Oracle, Alibaba, IBM, and smaller providers |
AWS remains the largest cloud infrastructure provider by revenue, holding roughly 29% to 30% of the global market as of early 2026, per market-share estimates cited in industry coverage. Microsoft Azure trails at around 25%, still growing faster than AWS on a percentage basis thanks to its OpenAI relationship and Copilot bundling. Google Cloud sits third at roughly 13%, the smallest of the three major providers but the fastest-accelerating in recent quarters.
The megadeal pattern helps explain why AWS’s share hasn’t eroded faster despite Azure’s momentum. Landing $6 billion from Snowflake and $4 billion from Pinterest in the same six-week window doesn’t show up as new market share overnight, but it locks in years of future consumption that a rival cloud provider would otherwise have a real shot at winning.
A Decade on AWS: How Snowflake Got Here
Snowflake was founded in 2012 by Benoit Dageville, Thierry Cruanes, and Marcin Zukowski, who built the company around a data warehouse designed from the ground up for the cloud rather than adapted from older on-premises software. It became generally available running on AWS in 2015, years before it added support for Microsoft Azure and Google Cloud.
The company went public on the New York Stock Exchange on September 16, 2020, under ticker SNOW, pricing shares at $120 and opening at $245 on day one, the largest software IPO at the time. Berkshire Hathaway and Salesforce both took stakes around the listing, an unusual move for Warren Buffett’s famously tech-averse firm that drew nearly as much coverage as the IPO’s size. Almost six years later, the AWS relationship that started as a hosting arrangement has grown into a $6 billion strategic commitment. Snowflake now reports more than $7 billion in lifetime sales through AWS Marketplace alone, a channel that generated over $2 billion in calendar 2025.
Snowflake vs. Databricks: The Multi-Cloud Question
Snowflake’s closest competitor, Databricks, has built its business on the opposite pitch: run anywhere, and avoid leaning too hard on any single hyperscaler. Databricks is privately held and hasn’t disclosed 2026 valuation or revenue figures, so a direct financial comparison isn’t possible from public data right now. The strategic contrast is visible anyway in how each company is positioning itself this year. Snowflake, while it still technically supports Azure and Google Cloud, just made its deepest and most public bet yet on a single provider. Databricks has continued marketing multi-cloud flexibility as a selling point against exactly that kind of concentration.
Which approach wins probably depends on the customer. Enterprises already committed to AWS at the platform level have every incentive to consolidate spend and collect the pricing benefits that come with it. Enterprises running genuinely multi-cloud environments, which describes most large enterprises, may see Snowflake’s AWS tilt as a reason to keep evaluating alternatives rather than deepen a single-vendor relationship of their own.
What AWS Gets Out of the Deal
The commercial logic runs in both directions. AWS Marketplace takes a cut of transactions that flow through it, and every dollar of Snowflake spend that AWS can point to as Marketplace revenue strengthens the pitch it makes to other software vendors to sell the same way. AWS also gets a marquee reference customer for Graviton at a moment when it’s trying to convince the broader market that custom silicon, not only Nvidia GPUs, is a viable path to cheaper AI infrastructure. OpenAI’s own move onto AWS Bedrock earlier in 2026 gave AWS a similar proof point with a very different kind of customer.
There’s a defensive angle too. Every year Snowflake’s compute runs deeper into Graviton and AWS’s AI stack is a year Snowflake becomes more expensive to migrate away from. That kind of switching-cost dynamic rarely gets stated outright in a press release, but it’s a standard part of how hyperscalers structure strategic collaboration agreements with their biggest software partners.
The Risk Nobody’s Pricing In Yet
Five-year, nine-figure-a-year infrastructure commitments only look smart in hindsight if the demand actually shows up. Snowflake is effectively betting that AI-driven data consumption keeps compounding at something close to its current 33%-34% growth rate for half a decade, a pace few software companies of Snowflake’s size have ever sustained that long.
If enterprise AI spending cools, plateaus, or shifts toward an architecture AWS doesn’t win, Snowflake is still on the hook for a $6 billion commitment sized for the growth case, not the slowdown case. The same risk applies in miniature to Pinterest and in a much larger form to Anthropic, whose $100 billion, decade-long AWS commitment assumes frontier AI model demand keeps climbing at a pace that has already proven difficult to forecast even one year out, let alone ten.
What It Means for Enterprise Cloud Buyers
For companies evaluating Snowflake, Databricks, or a direct AWS-native data stack, the practical takeaway is pricing leverage. Snowflake’s deeper Graviton commitment should, over time, translate into better price-performance for customers running large workloads, since AWS has consistently priced Graviton instances below comparable x86 alternatives.
The deal is also a signal worth watching for procurement teams. Hyperscalers are increasingly willing to trade steep multi-year discounts for guaranteed consumption, which means large enterprises with predictable workloads have more negotiating room right now than they did two years ago, when GPU and compute capacity was scarce enough that vendors, not buyers, set the terms.
Five Predictions for the Rest of 2026
- Expect at least one more billion-dollar-plus AWS infrastructure commitment from a named enterprise software vendor before the end of 2026, continuing the pattern set by Snowflake and Pinterest.
- Databricks will face growing pressure to make its own public infrastructure commitment, whether to AWS, Azure, or Google Cloud, to counter the narrative that Snowflake is winning the AI-platform positioning fight.
- Snowflake’s net revenue retention rate becomes the single most-watched metric in its next two quarterly reports, since any decline would undercut the growth assumptions behind the AWS deal.
- Microsoft and Google each expand their own custom-silicon partnerships, similar in spirit to AWS’s Graviton push with Snowflake, to keep pace on price-performance messaging.
- FinOps and cloud cost-governance vendors see rising demand from enterprises trying to model the long-term cost exposure of multi-year AI infrastructure commitments like this one.
That last prediction is already playing out. AWS’s own free FinOps agent, launched into a market estimated at $16.5 billion, and separate research showing cloud waste running as high as 29% of AI-era budgets both point to the same tension. Companies want the AI capacity badly enough to sign decade-long checks, and they need better tools to make sure that spending doesn’t spiral once the contracts are signed.
Frequently Asked Questions
How much did Snowflake commit to spend with AWS?
Snowflake committed to $6 billion in spending with AWS over five years, announced May 27, 2026, under an expanded Strategic Collaboration Agreement covering compute, AI integrations, and AWS Marketplace go-to-market.
Why did Snowflake announce the AWS deal the same day as earnings?
Snowflake paired the AWS commitment with its fiscal first-quarter 2027 results, which showed 33% revenue growth to $1.39 billion. Announcing both on the same day let the company present the earnings beat and the AWS spending as evidence of one story: accelerating enterprise AI demand.
How much did Snowflake’s stock rise after the announcement?
CNBC reported that Snowflake shares rallied as much as 36% following the combined earnings beat and AWS news on May 27, 2026.
What is AWS Graviton and why does it matter to this deal?
Graviton is AWS’s custom, ARM-based processor line, built to offer better price-performance than standard x86 chips from Intel and AMD. Snowflake is expanding its use of Graviton to lower its own compute costs, which supports margins without raising customer prices.
Is Snowflake leaving Microsoft Azure or Google Cloud?
No. Snowflake still supports Azure and Google Cloud. The AWS deal deepens one relationship without publicly ending the other two, though it does signal where Snowflake sees its biggest near-term growth.
How does this compare to other 2026 AWS infrastructure deals?
It’s the second-largest disclosed AWS commitment of 2026 behind Anthropic’s, and one of at least four megadeals AWS closed with major customers this year, alongside Pinterest and Meta.
What is net revenue retention, and why is Snowflake’s 126% figure significant?
Net revenue retention measures how much more, or less, existing customers spend year over year. A 126% rate means the average Snowflake customer increased spending by 26% without counting any new customers added during the period, a sign of deepening usage rather than just new sales.
Does the deal change pricing for existing Snowflake customers?
Snowflake hasn’t announced customer-facing pricing changes tied to the deal. Analysts expect any benefit to show up gradually through improved margins rather than direct price cuts in the near term.
