Data + AI Summit 2026: The Complete Guide
If you work anywhere near data engineering, machine learning, or AI, there's a decent chance you've heard people talking about Data + AI Summit. The 2026 edition already happened — it ran June 15–18 at the Moscone Centre in San Francisco, and it was, by most accounts, the biggest one yet. Roughly 31,000 people showed up in person, with tens of thousands more tuning in online from over 150 countries.
This guide explains what the event is, what happened this year, and why it matters whether you attended, watched from home, or are deciding if it's worth your time and budget next year.
A quick note on sourcing: this isn't written from firsthand attendance. It's pulled together from Databricks' own announcements and press materials, plus reporting from analysts, sponsors, and attendees who published recaps. Where something isn't officially confirmed or is specific to a future edition, I've said so rather than guessing.
Table of Contents
- Making the Networking Actually Pay Off
- Is It Worth It for Your Career?
- Should You Attend the Next One?
- Watch the Keynote
- What Is Data + AI Summit, Exactly?
- When and Where Did It Happen?
- Who Runs It
- Why People Actually Attend
- Registration and Ticket Prices
- What Was Actually on the Agenda
- The Big Themes
- The Announcements, Explained Simply
- What Changed From Last Year
- Workshops, Training, and Certification
- Should You Go In Person or Watch Virtually?
- Planning the Trip
- Frequently Asked Questions
- Related Reading
- Final Thoughts
What Is Data + AI Summit, Exactly?
Data + AI Summit is Databricks' annual flagship conference. The company calls it the world's largest data, analytics, and AI event, and that claim is hard to argue with. It started years ago as a technical, Apache Spark-focused user conference. It has since grown into a four-day showcase covering the modern data and AI stack, from pipelines and governance to business intelligence and, increasingly, AI agents built on enterprise data.
Databricks is the company behind the "lakehouse" — a way to unify data engineering, warehousing, and machine learning on a single platform rather than stitching together separate systems. More than 20,000 organisations reportedly run on Databricks in some form, including adidas, AT&T, Mastercard, Rivian, and Unilever, as well as roughly 70% of the Fortune 500. When a company with that kind of enterprise AI footprint throws its biggest event of the year, the sessions tend to reflect what large organisations are actually building right now, not just what's interesting in a research lab somewhere.
The audience matches that scope: data engineers, data scientists, ML engineers, analysts, application developers, and the technology leaders who fund all of the above. That mix matters because it shows the summit is built for both the people doing the work and the people deciding what to invest in.
<a id="dates-venue"></a>
When and Where Did It Happen?
Data + AI Summit 2026 ran June 15–18, 2026, across the Moscone Centre's North, West, and South buildings in San Francisco (747 Howard Street, if you're mapping it out). It was a hybrid event — in person plus a free virtual livestream — and it ran on Pacific Time throughout.
Worth knowing if you're planning a future edition: San Francisco hosted FIFA World Cup matches that same summer, and hotel prices near the venue spiked hard as a result. Check what else is happening in the city before you lock in your travel plans. Also, don't assume next year's dates will match this year's exactly — Databricks confirms them on the official Data + AI Summit page well in advance, and that's the source to check rather than last year's calendar.
<a id="organizer"></a>
Who Runs It
Databricks organises and runs the whole thing. If you're not familiar with the company, it builds the Lakehouse Platform, and its fingerprints are on a lot of the open-source tools you've probably heard of — Apache Spark, Delta Lake, MLflow. The summit is its biggest annual stage for product announcements, customer stories, and open-source community news.
It's not purely a Databricks infomercial, though. This year's program included a genuinely interesting fireside chat between OpenAI co-founder and president Greg Brockman and Databricks VP Patrick Wendell, as well as sessions on Anthropic, LangChain, LlamaIndex, Cognition, CrewAI, Glean, Lovable, and Replit. The AI agent-building conversation right now spans many companies, and this year's agenda reflected that.
<a id="why-attend"></a>
Why People Actually Attend
Strip away the marketing language, and the reasons people show up are pretty practical:
The learning is dense. 800+ breakout sessions across four days mean there's genuinely something for almost every skill level — from "what is a lakehouse" basics to deep, technical sessions on agent observability and cost control.
You hear the announcements first. Databricks tends to save its biggest platform news for the keynote stage — this year, that meant Genie One reaching general availability and the debut of Lakehouse//RT.
Certification is on offer. Onsite instructor-led training and certification exams mean you can walk away with more than notes — you get a credential. For readers, that's a concrete way to turn attendance into something you can use afterwards.
The networking density is hard to replicate. Tens of thousands of people working on similar problems, in one place, for four days straight. That gives readers a better chance to find peers, compare approaches, and make contacts that can outlast the event.
It's a visible career signal. Attending, speaking, or getting certified is a concrete way to show you're current on the tools employers are actively hiring for.
If your goal is deep, hands-on learning and real networking, in-person is worth it. If you mainly want to know what got announced, the free virtual stream covers most of that value without the travel bill. Choose the format that best matches the result you want. Either way, those tradeoffs lead directly into registration and pricing.
<a id="registration"></a>
Registration and Ticket Prices
Data + AI Summit 2026 had three tiers:
There was a 50% early-bird discount running from February 18 through April 30, 2026 — which is roughly half off if you register in time. If you're a student, Databricks doesn't seem to have a single, publicly posted discount code; check the official FAQ page or reach out directly, since policies here can shift year to year. The same goes for cancellations — there's no one blanket policy across all channels, so whatever's shown at checkout on the official registration page is what to trust.
Two things worth planning around: workshops and certification exams have sometimes needed separate sign-up beyond your general pass, so don't assume a Full Conference Pass automatically reserves you a seat in a specific hands-on session. And if you want the in-person experience, registering during the early-bird window is the simplest way to cut the cost roughly in half.
<a id="agenda"></a>
What Was Actually on the Agenda
The keynote program leaned heavily on Databricks leadership and outside AI labs. CEO Ali Ghodsi opened the event with a framing that echoed through nearly every session afterwards: AI's real limitation right now isn't intelligence; it's context. If a CFO can't get an AI system to explain why margins moved, that's a context gap, not a model problem. CTO Matei Zaharia picked up a related thread on costs — agentic AI usage is getting genuinely expensive — and he discussed the need for smarter layers on top of models and agents to keep spending under control. Co-founder Reynold Xin took the stage to introduce Lakehouse//RT, a new real-time analytics engine, and the Brockman/Wendell fireside chat dug into what it actually takes to connect AI models to real business workflows rather than just chat interfaces.
Customer stories referenced on stage included PepsiCo, Mastercard, and AstraZeneca, among others. Beyond the main stage, there was the usual social programming: a Welcome Reception, 80+ smaller special-interest meetups, an annual Women in Data + AI gathering, and the signature evening party, "Data After Hours," held this year at Oracle Park with a live musical act.
<a id="themes"></a>
The Big Themes
Databricks framed the whole event around four pillars: Context, Control, Cost, and Choice. In practice, that broke down into a few recognisable clusters of content:
Agents and generative AI. Genie One, Genie Ontology, Agent Bricks, and Omnigent were the headline names here — tools aimed at building AI agents and grounding them in real business context rather than letting them guess.
Data engineering. Continued work toward Delta Lake and Apache Iceberg interoperability, updates to Lakeflow for pipeline orchestration, and the new Lakehouse//RT engine for real-time workloads.
Governance and security. Unity Catalogue Metrics, the Unity AI Gateway, and Catalogue Federation — essentially, extending the governance rules that have always applied to tables and dashboards so they also apply to agents and the tools those agents use.
Analytics and BI. Databricks SQL, the AI/BI Genie interface, and Lakebase, a Postgres-compatible database designed to power AI applications that require fast, transactional data access.
Open ecosystem. Sessions touching LangChain, LlamaIndex, DSPy, dbt, Trino, and the Model Context Protocol (MCP) rounded things out.
If you only have time to pick a handful of sessions from any given year's catalogue, it's worth matching them to whichever cluster maps onto your actual job rather than trying to cover everything.
<a id="announcements"></a>
The Announcements, Explained Simply
A handful of names came up constantly throughout the week, so here's what they actually are:
Genie One reached general availability this year. It's Databricks' conversational, agentic interface — the idea is that business teams can ask questions in plain language and get real answers grounded in company data, not just a chatbot guessing.
Genie Ontology is the context layer underneath Genie. Databricks describes it as self-improving — it automatically pulls in and updates business knowledge from Databricks and connected workplace apps, so Genie's answers are grounded in governed data rather than static documents someone wrote six months ago.
Agent Bricks and Omnigent are the tools for actually building and managing agents, built so they automatically respect the same permission and masking rules that already govern your data in Unity Catalogue.
The Unity AI Gateway sits between agents and the rest of your systems. One of its more interesting features is smart routing — it can automatically send easy tasks to cheaper models and harder ones to more capable (and pricier) models, which is a direct response to how expensive agent usage can get if left unmanaged.
Lakehouse//RT is a new SQL warehouse type built for real-time, high-concurrency workloads, running directly against your existing Delta Lake or Iceberg tables — no separate serving layer, no duplicated data.
Lakebase is a serverless, Postgres-compatible database designed for AI applications and agents that need fast, transactional data access, with branching features that let teams spin up isolated test environments without copying production data.
Nearly every announcement circled back to the same argument from the keynote stage: today's AI models are already capable enough. The real bottleneck is providing them with trustworthy context, governed as data has always been.
<a id="changed"></a>
What Changed From Last Year
DAIS 2025 was mostly about introducing new building blocks — Agent Bricks, Lakebase, Lakeflow, and expanded Mosaic AI capabilities all made their debut. 2026 was less about new primitives and more about proving those primitives could actually run in production at enterprise scale, with governance and cost control built in from the start.
If last year was about "look what's possible," this year was about "here's how to actually run this without your AI bill spiralling or your compliance team panicking."
<a id="training"></a>
Workshops, Training, and Certification
Beyond the keynotes, there were 25+ hands-on training and certification courses this year, spanning everything from Lakebase and Genie to AI-assisted development. There was also a multi-day hackathon run in partnership with OpenAI, focused on building agentic data apps for social impact.
Training generally spans three tiers:
Beginner: lakehouse fundamentals, basic Delta Lake data engineering, foundational AI/ML concepts
Intermediate: Unity Catalogue governance setup, Databricks SQL for analytics teams, MLOps workflows
Advanced: performance tuning, agent architecture and observability at scale, multi-cloud governance
If certification matters to you, double-check whether the specific course requires its own registration — that's historically been separate from the general conference pass.
<a id="virtual-vs-in-person"></a>
Should You Go In Person or Watch Virtually?
This is really a question about what you need out of the event.
The free virtual track is a genuinely solid option if you just want to stay current — it's one of the few major tech conferences that doesn't paywall its keynote content. But hands-on skills and a certification credential are available only with an in-person or full-access pass.
<a id="travel"></a>
Planning the Trip
If you do go in person, a bit of planning goes a long way. SFO is the closest major airport — about 30 to 45 minutes from downtown by BART, car, or rideshare. Oakland and San Jose airports work as backups if SFO fares get expensive. Once you're downtown, Moscone Centre is walkable from most Union Square and SoMa hotels, and Databricks has historically partnered with nearby hotels for group rates through its own registration page — worth booking early, especially in a year where San Francisco has other major events overlapping.
Realistically, budget somewhere in the $2,500–$4,000+ range all-in for ticket, flight, hotel, and food if you're travelling from elsewhere — obviously that swings a lot depending on your departure city and how far ahead you book.
<a id="networking"></a>
Making the Networking Actually Pay Off
The people who get the most out of the event tend to treat networking as something you plan for, not something that just happens. A few things that help: update your LinkedIn before you arrive, have a short, specific way of describing what you work on, carry a quick way to share contact info, and actually follow up with people within a few days while the conversation is still fresh in both your heads. It's also worth resisting the urge to pack every keynote into your schedule — the smaller special-interest meetups are often where the more useful, specific conversations happen.
<a id="career"></a>
Is It Worth It for Your Career?
If a job search or career growth is part of why you're considering this, there are a few concrete angles: the skills taught track closely with what companies are actually hiring for right now (agent development, governance, lakehouse architecture), the certification is a real, resume-visible credential, and recruiters are physically present at sponsor booths, not just at a virtual job board. Bring an updated resume if that's your goal, and don't treat sponsor booths as just a place to grab a t-shirt — they're a legitimate networking channel.
<a id="next-edition"></a>
Should You Attend the Next One?
It depends on your role, honestly. If you're actively building AI agents on enterprise data, or you own governance and compliance, there's a strong case for going in person — those were the two biggest themes this year, and that's unlikely to change soon. If you're earlier in your career or just exploring the space, the free virtual track plus a few introductory sessions probably cover most of what you'd get out of it. And if your organisation doesn't use Databricks and has no plans to, the ideas are still broadly relevant to data and AI strategy, but a more vendor-neutral event might map better onto your actual stack.
<a id="watch"></a>
Watch the Keynote
Databricks posted the full keynote recordings on its YouTube channel if you want to watch rather than read a recap:
Data + AI Summit Keynote 2026 — Day 1 — Ali Ghodsi's opening, the "context problem" framing, Unity AI Gateway, Genie Ontology, and Genie One announcements
Data + AI Summit Keynote 2026 — Day 2 — further product deep-dives from the Databricks engineering team
If you're embedding this on a website, a standard YouTube iframe embed works directly with either video ID above (e.g., <iframe src="https://www.youtube.com/embed/Qux8E-L1mk8">).
<a id="faq"></a>
Frequently Asked Questions
What is Data + AI Summit 2026?
Databricks' annual flagship conference for data engineering, analytics, and AI, held June 15–18, 2026, at the Moscone Centre in San Francisco, with roughly 31,000 in-person attendees and a large virtual audience.
Who organises it?
Databricks — the company behind the Lakehouse Platform and open-source projects like Apache Spark, Delta Lake, and MLflow.
Where is it held?
The Moscone Centre (North, West, and South buildings) in San Francisco.
When did it run?
June 15–18, 2026, Pacific Time.
Can I attend online?
Yes, for free — a Virtual Pass gets you livestreamed keynotes and a curated set of sessions, with some available on-demand afterwards.
How much does it cost?
Roughly $1,895 for the Full Conference Pass (with a 50% early-bird discount available earlier in the year), about $195 for Keynote & Expo only, or free for virtual.
Is there a student discount?
Not a universally published one — check the official FAQ or contact Databricks directly.
Who should attend?
Data engineers, data scientists, ML/AI engineers, analytics professionals, platform architects, and technology decision-makers are included in the audience, though the agenda also includes beginner content.
Are workshops and certification included in my ticket?
Sometimes they require separate sign-up — check the specific session listing rather than assuming.
What should I bring?
Laptop, charger, a portable battery pack, a way to share contact info, comfortable shoes, and a resume if you're job-hunting.
Are sessions recorded?
Yes, keynotes and a range of sessions are typically posted on-demand afterwards on the official Databricks site.
What was this year's theme?
"Apps and agents that work" — the idea that agentic AI is moving out of the lab and into governed, cost-aware production systems.
What's the biggest thing to know about the announcements?
Genie One hit general availability, Genie Ontology launched as its context layer, and Lakehouse//RT introduced real-time analytics directly on existing Delta Lake or Iceberg tables — all tied to the idea that AI's real bottleneck is business context, not model quality.
Will there be a next edition?
Very likely, given the event's history, but confirm dates on the official Data + AI Summit page once they're announced rather than assuming they'll match this year's.
<a id="related"></a>
Related Reading
• Is Character AI Down? Live Status & Quick Fixes
• How Much Water Does AI Use Per Prompt? Full Environmental Breakdown
• How Much Water Does AI Use? The Real Cost of Every AI Question?
• Game AI Explained: Techniques, Examples & Tools
• What Is an AI Automation Agency? A Complete Guide (2026)
Final Thoughts
If you zoom out, Data + AI Summit 2026 reflects a shift that's been building for a couple of years: the conversation has moved from "here's how to manage your data" to "here's how to build AI agents you can actually trust on top of the data you already have." The "Context, Control, Cost, Choice" framing captured that well — agentic AI moving from experimentation into governed, cost-aware production, with governance now stretching to cover agents and tools, not just tables and dashboards.
If you were there, the most useful thing to do now is simple: go back through your notes on the two or three sessions that were actually relevant to your job, and turn them into something concrete at work in the next month — a new pipeline pattern, a governance policy, an agent prototype — while it's still fresh.
If you weren't there, the free recordings are a legitimate way to catch up on the announcements without paying for a pass, and registering interest early for the next edition is the best way to catch early-bird pricing if you decide you want the in-person experience next time.
Check the official Data + AI Summit page for the latest confirmed details on any future edition — dates, pricing, and policies can all shift year to year, and that's always the source to trust over any recap, including this one.
FAQ Schema (JSON-LD)
{
"@context": "https://schema.org",
"@graph": [
{
"@type": "Article",
"headline": "Data + AI Summit 2026: The Complete Guide",
"datePublished": "2026-07-05",
"dateModified": "2026-07-05",
"author": { "@type": "Person", "name": "Your Author Name", "url": "https://example.com/author/your-author-name" },
"publisher": {
"@type": "Organization",
"name": "Your Site Name",
"logo": { "@type": "ImageObject", "url": "https://example.com/logo.png" }
},
"mainEntityOfPage": "https://example.com/data-and-ai-summit-2026-guide"
},
{
"@type": "BreadcrumbList",
"itemListElement": [
{ "@type": "ListItem", "position": 1, "name": "Home", "item": "https://example.com/" },
{ "@type": "ListItem", "position": 2, "name": "AI Conferences", "item": "https://example.com/ai-conferences" },
{ "@type": "ListItem", "position": 3, "name": "Data + AI Summit 2026 Guide", "item": "https://example.com/data-and-ai-summit-2026-guide" }
]
},
{
"@type": "Organization",
"name": "Databricks",
"url": "https://www.databricks.com",
"sameAs": ["https://www.linkedin.com/company/databricks", "https://www.youtube.com/c/Databricks"]
},
{
"@type": "Person",
"name": "Your Author Name",
"url": "https://example.com/author/your-author-name",
"jobTitle": "AI & Data Industry Writer"
},
{
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "What is Data + AI Summit 2026?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Databricks' annual flagship conference for data engineering, analytics, and AI, held June 15-18, 2026 at the Moscone Center in San Francisco."
}
},
{
"@type": "Question",
"name": "How much does it cost to attend?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Roughly $1,895 for the Full Conference Pass (with a 50% early-bird discount available earlier in the year), about $195 for Keynote & Expo only, or free for virtual attendance."
}
}
]
}
]
}
0 Comments