At its annual Data+AI Summit, Databricks is highlighting both Genie and a series of new capabilities to get enterprise data ready for AI agents. In doing so, Databricks is essentially telling the same story as Snowflake, SAP, and Salesforce, but the company hopes to differentiate itself with a consistent open-source strategy. That story is finding a receptive audience in Europe.
At its Data+AI Summit, Databricks has made a series of announcements that all fit into the same broader narrative: preparing enterprise data for AI. CEO Ali Ghodsi summarizes the mission succinctly: “AI doesn’t have an intelligence problem, at least not in the enterprise. AI has a context problem.” According to him, the models are smart enough. The problem is that they don’t know the data, the context, and the processes within organizations.
The common thread between the announcements is always the same. Databricks wants to bridge the gap between raw enterprise data on the one hand and AI agents that need to do something meaningful with it on the other.
A context genie
Genie is the flagship offering in this regard. With Genie One, business users can ask questions of their data in plain language, while Genie Ontology builds a knowledge graph of all contextual knowledge in an organization in the background.
Ghodsi compares this to what PageRank once did for Google: instead of an agent trudging through documents for minutes, the system already knows in advance which data is relevant. Just like a human, an AI system will be able to interpret questions within the correct business context.
Beneath Genie lies a renewed data foundation. With Lakehouse RT, Databricks promises an engine that is ten to a hundred times faster in terms of latency, which is crucial for agents that need to query data quickly. And with LTAP, the company is bringing transactional databases and classic data warehousing together in one layer for the first time. According to Ghodsi himself, this is something the industry has been waiting for for forty years.
Everyone is telling the same story
Anyone who has attended multiple tech conferences this year, heard a very familiar sound at Databricks. The story sounds almost identical at the competition: “Data is the fuel, we have your data, agents run on it, we provide the much-needed context layer and governance.”
At arch-rival Snowflake, everything equally revolves around the idea that data is the foundation of every AI transformation, with a strong emphasis on openness via the Iceberg format and a cloud-agnostic approach. The story at the Snowflake Summit, barely two weeks before Databricks’ in the same Moscone Center in San Francisco, showed many similarities. The two companies are visibly moving toward each other.
The same message resonates outside the classic data world as well. Earlier this year at Sapphire, SAP outlined its vision for the autonomous enterprise, in which AI agents work with enterprise data and context is again the keyword. Even Celonis tells its version of the same story.
If everyone claims the rights to the same central role, why should a customer choose Databricks?
Can openness make the difference?
The answer Databricks gives is consistent: open source and no vendor lock-in. Before the summit, the company transferred its Open Sharing protocol to the Linux Foundation, in line with the open-source roots the company says it has had since day one.
Samuel Bonamigo, senior vice president and general manager EMEA, emphasizes that this philosophy is not marketing, but a real differentiator. The data always belongs to the customer, is the core of his argument, and thanks to the open nature, customers can also take their data with them. At the Summit, we notice that Databricks, as a company not yet listed on the stock exchange, is strongly supported by technical profiles who truly believe in openness.
Ghodsi also hammers home the importance of that openness, but from a broader, almost societal perspective. “I think it would be a big risk if open source were banned,” he states, in this case referring to AI models. “As long as open source exists as an alternative, universities and researchers worldwide can look inside the technology. If that disappears, only the handful of frontier companies remain that are still allowed to look under the hood.”
“I think it would be a big risk if open source were banned”
Ali Ghodsi, CEO Databricks
Databricks explicitly positions itself as the party that offers customers choice and flexibility, in a market where the best model sometimes only stays at the top for a month.
That openness is more than a principled stance. It translates concretely into the ability to switch models, decouple storage and compute to control costs, and store data in an open format that is not tied to Databricks. According to Bonamigo, that combination attracts European customers, although we add for the sake of honesty that model openness is also offered by the other players.
Open, but not practical
Yet there is a catch to all that openness. Databricks preaches “no lock-in,” but acknowledges itself that managing pure open source rarely works in practice. Rich Radley, VP Field Engineering EMEA, admits that even large customers with substantial budgets and strong engineering teams try, but fail to maintain quality when running open-source Spark on Kubernetes with other open-source frameworks.
“Most customers fail at that once they reach a certain scale,” he acknowledges. Radley points to the return on investment: “the size of the engineering and SRE teams that such an open-source infrastructure requires can also be spent on something else by simply using Databricks’ managed service.” In other words: the openness remains real, but in practice, it still leads the customer to Databricks’ own paid platform.
A seat at the table
This brings us to the local dimension, because the open-source strategy gives Databricks a place at the European table that competitors don’t easily have. Glenn Lottering, ACP and TechGM Field Engineering for Benelux and Nordics, states that the open nature literally earns Databricks a seat in the forums where the future of AI in the EU is discussed. “Because we are open source, we get a seat at the table,” he beams.
“Because we are open source, we get a seat at the table.”
Glenn Lottering, ACP and TechGM Field Engineering Benelux and Nordics, Databricks
The company is closely monitoring European movements around sovereignty and says it could adapt if the European market evolves toward its own cloud, thanks to its cloud-agnostic nature.
Open source, sovereignty, and data privacy are more sensitive in the EU than elsewhere, and that plays to Databricks’ advantage. Lottering notes that most customers cite openness as one of their deciding factors, especially in the current geopolitical context. Bonamigo adds that management with the open Unity Catalog is, in his view, a unique capability that also continues to evolve with customer needs. Yet we hear echoes here of what Snowflake says about its (less open) Horizon Catalog.
Footprint in the Benelux
Furthermore, the Benelux is not a side issue for Databricks. Regional VP Benelux Kevin Jonkergouw points out that the company’s first EMEA R&D hub was established in the region and is still the largest.
“The software we roll out globally is also built in the Benelux.”
Kevin Jonkergouw, Regional VP Benelux Databricks
“More than four hundred people are now working there, and the ambition is to reach a thousand employees in the coming years,” says Jonkergouw. “The software we roll out globally, including at the Summit, is also built in the Benelux. Customers in the Benelux and the Nordics are also so intertwined with the technology that they help develop the product.”
From data player to AI specialist, if the customer is on board
There is a caveat to the story, and it applies equally to the competition: Databricks did not start as an AI company. The company originated around Spark and data processing and is now positioning itself as an AI specialist. Ghodsi himself points out that data and AI have always gone hand in hand for Databricks, but the shift to a full-fledged enterprise AI platform is real. That only succeeds if customers buy into the story.
At Heineken, that adoption is not automatic. Jelle van Etten, Head of Global Data Platform, explains that the beer company standardizes its markets via templates and Genie skills, so that local teams can build use cases themselves.
He is honest about the pitfalls: “In the beginning, we assumed that every engineer would want something like this and would immediately know how to use it,” he says. In practice, Heineken must continue to convince, train, and set up communities to get people to use the tools in the right way. In other words, AI adoption doesn’t happen by itself.
At Amadeus, active in the highly fragmented travel sector, they see similar patterns. Abhishek Krishna, Head of Data, AI & Platform, says the company appoints champions within each team to drive change, and that trust in the data is the key to adoption.
The concrete result is impressive: for the customer success teams, the time to compile a report dropped from an average of fourteen days to three minutes. His colleague Michael Yeomans, SVP Travel Intelligence, emphasizes with a nice sector-related pun that it is a journey and not a destination, and that the human factor remains crucial: “the intention is to support employees, not replace them.”
For Amadeus, it is precisely the multi-cloud and open nature of Databricks that is decisive, because customers are spread across multiple clouds and on-premises.
Central, but open, hub
So Databricks, like so many tech players, wants to play a pivotal role in its customers’ AI transition. The ingredients it puts on the table (the fast data layer, a context layer via Genie, management, and agents) differ essentially little from what Snowflake, SAP, or Salesforce promise. There is a broad consensus on what companies need: the battle is over who delivers on that promise most credibly.
Databricks is betting everything on one trump card: openness. The open format, the cloud-agnostic approach, and the consumption model where the company says it only earns when the customer actually gets value, are intended to make the difference. The fact that this openness runs into practical limitations at large enterprises doesn’t matter that much: for customers, it’s mainly about not wanting to be truly locked in.
For Europe, that is more than a technical choice. The open-source focus earns Databricks a seat at the table where sovereignty and data privacy are at the top of the agenda. Whether that open-source trump card is enough to permanently differentiate itself from the competition will have to be seen in the coming years.