Snowflake is about to spend $250 million on Crunchy Data, which generates just over $30 million annually. Databricks, meanwhile, dropped about $1 billion on Neon—probably with even less revenue—followed by a $1 billion–$2 billion deal for Tabulr simply to keep it out of Snowflake’s hands. Both companies claim these buys are strategic for “agentic AI,” but the underlying reality is much more cynical. These multibillion-dollar checks serve as a brutal display of one-upmanship that punishes common sense.
At first glance, Snowflake’s $250 million acquisition of Crunchy Data seems reasonable. Crunchy Data provides a hardened, cloud-hosted PostgreSQL distribution with security and performance tweaks. Snowflake can integrate it into a managed service and channel more data into its analytics engine. Crunchy Data brings around 100 employees, a loyal customer base, and an open-source pedigree.
Yet Snowflake passed on buying Neon last year—a startup with similar PostgreSQL roots but boasting serverless compute, branching, and time-travel features that align perfectly with Databricks’ AI-centric vision. Instead of grabbing Neon, Snowflake pivoted to Crunchy Data, suggesting it cares less about pure technology and more about sending a market signal: “We’ll still play the database game even if we missed Neon.” The lesson? Often these acquisitions are driven by ego and optics rather than by product-fit or customer value.
Databricks never pretends to flinch at high price tags. It shelled out roughly $1 billion for Neon—plausibly over thirty times Neon’s annual revenue. Then it outbid Snowflake for Tabulr at up to $2 billion. Tabulr’s pitch revolved around a code-centric interface that fused data prep with AI suggestions—ideal for a future built on generative AI. Databricks couldn’t risk Snowflake getting that edge, so it paid a sky-high premium.
Databricks then wrote a $1.3 billion check for MosaicML—an LLM vendor with vague revenue numbers. The message was clear: if a startup so much as whispers that Snowflake is interested, Databricks will swoop in with a fire sale bid. This behavior creates a perverse incentive for startups to inflate valuations by hinting at interest from either giant. Founders gladly cash out rather than navigate the uphill battle of scaling organically.
Every analyst can point out how irrational these valuations are. Why do Snowflake and Databricks pay thirty-plus times revenue for startups? Shouldn’t a public company like Snowflake adhere to disciplined financial scrutiny? Snowflake will need to write down part of that $250 million if Crunchy Data fails to meet aggressive cross-sell targets. Databricks faces the same risk: if Neon’s serverless features don’t integrate seamlessly, those write-offs will hurt once it goes public or seeks another funding round.
Yet neither company seems concerned about economic rationale. By overpaying, they effectively bulldoze any chance for smaller competitors to acquire these assets. Startups with genuine innovation stay sidelined because VCs and founders chase exit prices inflated by this bidding war. The real casualties are customers and innovators who suffer from higher prices and fewer truly differentiated products.
Both Snowflake and Databricks claim these acquisitions will propel them into the era of agentic AI—autonomous systems executing complex tasks. But they’re doubling down on databases and data pipelines when the real goldrush lies elsewhere: AI assistants, vector databases, and integrated developer tooling. Microsoft, Google, Amazon, and Salesforce are racing to provide agent platforms, SDKs, and code assistants. Snowflake and Databricks, in contrast, keep snapping up database startups as if owning the storage layer alone will win the AI race.
The prime value in agentic AI comes from seamless integration of natural language understanding, custom model training, and edge deployment—areas neither Snowflake nor Databricks lead in. Their fixation on acquisitions reveals a deeper flaw: they’re trying to mask gaps in their AI offerings by buying market share instead of building differentiated products that address developer needs directly.
Take Tabulr: Databricks paid up to $2 billion just to keep it out of Snowflake’s hands. Tabulr’s unique proposition was a code-friendly interface that merged data preparation with AI insights—perfect for a world where generative AI generates and refines queries on the fly. If Snowflake had acquired Tabulr, it could have matched Databricks on AI-driven data workflows. Instead, Tabulr is now wrapped into Databricks’ larger platform, risking dilution of its standout features in a monolithic ecosystem.
That outcome hurts developers. They lose a focused, purpose-built tool and are forced into a sprawling platform that may prioritize other features. Rather than fostering competition and specialization, consolidation under a giant’s umbrella often leads to feature bloat and reduced innovation. End users pay more for watered-down versions of what could have been best-in-class products.
Crunchy Data’s appeal lies in its open-source roots and tight PostgreSQL integration. However, Snowflake will almost certainly wrap it behind premium pricing—charging for always-on clusters and advanced features. Crunchy Data has no serverless offering for OLTP today, unlike Neon. Snowflake can promote a native ingestion path from PostgreSQL into its analytics engine, but most customers will face separate charges for Crunchy Data’s premium add-ons.
If Snowflake genuinely wanted to champion open source, it could have backed Neon’s serverless model. Instead, it opted for Crunchy Data because its $250 million price tag fit Snowflake’s narrative of a hardened open-source core—one that the company can monetize heavily. This move undervalues true open-source collaboration and signals that Snowflake would rather monetize community-driven innovation than nurture it.
Perhaps the most controversial angle is how these acquisitions warp the startup ecosystem’s incentives. The promise of a near-term, billion-dollar exit forces founders to chase valuation multiples over product-market fit. A small startup solving a niche problem in AI agents or vector search won’t attract absurd bids unless it signals interest from Snowflake or Databricks. That dynamic sidelines countless innovators who might build more impactful solutions if they weren’t pressured to chase an inflated exit.
The focus shifts from customer-centric innovation to “who can outbid whom next.” Founders craft pitch decks designed to hint at interest from the two giants. VCs value companies not on revenues or product quality but on perceived strategic value to Snowflake/Databricks. It’s a vicious cycle where inflated valuations breed more inflated valuations—until the whole house of cards collapses under investor cold feet or disappointing integrations.
Critics on social media already point out the absurdity: Crunchy Data sold at roughly eight-times revenue, while Neon fetched thirty-plus times revenue. Private Databricks can act without shareholder approval; public Snowflake cannot. That disconnect means Snowflake will either overpay or walk away from targets it truly needs. When boards sign off on these deals, they must rationalize paying $250 million for a $30 million-revenue business. On what basis? Future cross-sell potential?
Even if Crunchy Data helps onboard more PostgreSQL users into Snowflake, the math remains difficult. Can Snowflake realistically extract enough incremental revenue to justify its outlay? Or is this a vanity play to show the market it still has firepower? Databricks, less accountable to public markets, can afford to write big checks. But when those acquisitions fail to deliver, the financial blowback will be severe—either in write-downs or in lost growth opportunities. As both companies jockey for narrative dominance, they risk undermining investor and customer trust.
This is classic procurement chicken. If Snowflake acquires Crunchy Data at $250 million, Databricks must secure a startup of equal or greater perceived value. If Databricks pays $1 billion for Neon, Snowflake chases Crunchy Data. Each move escalates the stakes. Neither side can back down without looking weak. That arms-race logic traps both companies into increasingly reckless spending. The ecosystem loses clarity and efficiency as hype and optics outweigh product roadmaps or customer value. Startups hold out for the highest bidder rather than focusing on building sustainable, differentiated offerings.
Proponents argue consolidation delivers fewer integrations and more seamless experiences. History suggests otherwise. When giants buy startups for inflated prices, the acquired teams often lose autonomy. Crunchy Data’s developers may become a support arm for Snowflake’s engineers, focusing on bug fixes rather than innovation. Neon’s serverless magic might be absorbed into Databricks’ lakehouse, diluting the feature set that originally made it special. Founders who once led product vision become cogs in a massive corporate machine. Customers end up paying premium subscriptions for solutions that never live up to the promise of standalone products.
If the goal is to accelerate agentic AI adoption, neither Snowflake nor Databricks is on the right path. Agents need real-time data integration, yes—but they also demand specialized infrastructure for natural language understanding, custom model training, and edge deployment. That won’t come from retrofitting acquired startups into sprawling platforms. Instead, fostering an open ecosystem of standards, developer-centric tooling, and transparent governance is vital.
Microsoft, Google, and Amazon invest heavily in open frameworks (e.g., ONNX, TensorFlow, SageMaker), sponsor academic research, and release SDKs that empower startups. Databricks and Snowflake, by contrast, seem fixated on buying market share. That strategy undermines community trust and slows real innovation. Paying $1 billion for Neon or $250 million for Crunchy Data is a vanity play. The real prize lies in enabling thousands of startups to build on flexible, open infrastructure—only then will agentic AI truly thrive.
These acquisitions raise uncomfortable questions for investors and customers alike. Why trust a platform that measures success by acquisitions rather than by empowering startups? How can a database layer bought at thirty-plus times revenue ever justify its price tag? When Snowflake and Databricks prioritize acquisition over organic growth, they expose a fundamental weakness: their core offerings may lack the innovation needed to sustain growth. Paying outrageous sums only masks that reality.
This is a desperate scramble to own the narrative and inflate market share. Until Snowflake and Databricks shift their focus toward genuine, community-driven innovation, their acquisition strategy remains a cautionary tale of how hype and ego can distort an entire industry.
📣 Want to advertise in AIM Research? Book here >
Cypher 2024
21-22 Nov 2024, Santa Clara Convention Center, CA
A Vendor Briefing is a research tool for our industry analysts, and an opportunity for a vendor to present its products, services and business strategies to analysts who cover the vendor specifically or a related technology or market.
AIM Research encourages technology vendors and agencies to brief our team for PeMa Quadrants, when introducing a new product, changing a business model, or forming a partnership, merger, or acquisition.
AIM Americas
2955, 1603 Capitol Avenue, Suite 413A, Cheyenne, WY, Laramie, US, 82001
AIM India
1st Floor, Sakti Statesman, Marathahalli – Sarjapur Outer Ring Rd, Green Glen Layout, Bellandur, Bengaluru – 560103
info@aimresearch.co
