Databricks CEO says enterprise AI remains stuck at the starting line

Databricks CEO: AGI Is Already Here, Why Companies Are Behind, and the Superintelligence Debate

The episode separates dramatic AGI claims from the slower, harder work of governing data, managing risk, and changing large organizations.

3 key takeaways
  1. 1Databricks’ early cloud, open-source, and machine-learning bets helped it build a durable data-and-AI platform.
  2. 2Ghodsi argues AGI may already meet older definitions, but enterprises still struggle with liability, security, cost, and context.
  3. 3AI’s economic impact will likely arrive gradually as organizations redesign workflows, management, and data governance over the next decade.

Don't miss

Ghodsi demonstrates Databricks Genie analyzing organizational data to expose meetings, decisions, costs, output, customer feedback, and unresolved management problems.

The brief

Databricks CEO Ali Ghodsi traces the company to Berkeley’s AMPLab, where limited industry interest in machine learning pushed researchers to build a business around making data and AI useful across organizations.

Ghodsi says AGI may already exist by older research definitions, but that milestone matters less than the enterprise gap: most companies remain limited to search, chatbots, and modest coding assistance.

The barriers are practical and intertwined: agents need internal context, yet organizations must control liability, cybersecurity, data sovereignty, and rapidly rising token costs before deploying them broadly.

In a Genie demonstration, Ghodsi shows how meetings, documents, communications, decisions, surveys, and operating data could give executives continuously updated evidence about how work gets done.

The conversation ends on a slower forecast: AI may improve the world, but deep economic transformation will take a decade or more as large organizations adapt.

What was said on this episode

49 statements · 24 positive · 12 negative · 4 mixed · 9 neutral

  1. Current AI provides enough intelligence to completely transform the planet.

    “Now I think we have all the intelligence we need to completely transform the planet.”

    Listen at 0:06

  2. Ali Ghodsion AI-driven economic transformationNegative0:28

    AI-driven economic transformation will take decades and GDP will move slowly.

    “But I actually think GDP will move very slowly and it's going to take decades.”

    Listen at 0:28

  3. AI writing 98% of code does not constitute recursive self-improvement.

    “Just because 98% of the code is being written by AI, we don't have RSI.”

    Listen at 0:36

  4. Databricks had a $7 billion run rate growing 80% year over year in August 2026.

    “August 2026, Ali, $7 billion run rate growing 80% year over year.”

    Listen at 1:49

  5. Ali Ghodsion organizational data science teamsPositive3:49

    Nearly every organization will need a data science team for AI work.

    “more organizations, probably, probably every organization will need a data science team that can do this kind of AI stuff.”

    Listen at 3:49

  6. Cloud computing separated storage from compute and decentralized data infrastructure.

    “What the cloud enabled was to just completely decentralize this and separate compute and storage”

    Listen at 6:03

  7. Ali Ghodsion DatabricksPositive6:52

    Databricks succeeded because it bet on cloud, AI, and open source.

    “the only reason we succeeded was because of these bets”

    Listen at 6:52

  8. Artificial general intelligence has already been achieved.

    “I actually think we already have artificial general intelligence.”

    Listen at 9:43

  9. Existing AI intelligence is sufficient to completely transform the planet.

    “we have all the intelligence we need to completely transform the planet.”

    Listen at 10:17

  10. Superintelligence could perform a millennium of human thought in one minute.

    “He's talking about something that can do a millennia worth of thought. So like a thousand years of human thought, it can do it like in a minute.”

    Listen at 10:57

  11. Ali Ghodsion enterprise AI adoptionNegative12:20

    Applying existing AI inside organizations is the biggest current challenge.

    “the biggest challenge that's facing us is how do we actually just make this AI applicable inside organizations?”

    Listen at 12:20

  12. Ali Ghodsion Enterprise AINegative12:33

    Enterprises mostly use AI as rudimentary search and chatbot tools.

    “they're just using AI in very rudimentary ways. They're just using it as a glorified Google search.”

    Listen at 12:33

  13. Ali Ghodsion AI coding toolsMixed12:49

    Enterprise AI coding tools currently deliver roughly 20–30% speedups.

    “with the coding... Actually, they're not seeing enough speed-ups. They're seeing maybe like 20%, 30%.”

    Listen at 12:49

  14. Ali Ghodsion enterprise autonomous agentsNegative13:04

    Enterprises have not yet deployed thousands of autonomous agents for core work.

    “It's just not happened yet.”

    Listen at 13:04

  15. Ali Ghodsion Enterprise AIPositive14:31

    AI could solve enterprise problems now if given complete organizational context.

    “If you capture all that context and you could feed it to the AI, they would nail the problems. They could help us right now already.”

    Listen at 14:31

  16. Ali Ghodsion AI token costsNegative15:41

    AI token costs are increasing exponentially as usage expands.

    “the token cost is going up exponentially.”

    Listen at 15:41

  17. Ali Ghodsion Enterprise AINegative15:47

    Companies have not seen revenue accelerate exponentially enough to match AI costs.

    “I don't know any company. who started using AI, and as of their use of AI, their revenue started accelerating exponentially, matching their cost curve.”

    Listen at 15:47

  18. Organizations gain substantial value from AI only after addressing four deployment barriers.

    “only then can you start getting something out of it.”

    Listen at 17:44

  19. Ali Ghodsion DatabricksPositive19:16

    Greater enterprise agent usage will increase Databricks query consumption.

    “all these companies, whenever they have agents and they're using it, then I think we're just in the beginning. As I said, I still think it's mostly Google search and coding, but the agents, when they start using that, they just send us more queries.”

    Listen at 19:16

  20. Ali Ghodsion organizational structureNeutral20:09

    Product sales and service delivery cut across hierarchical organizational charts.

    “But if you look at how actually any organization actually... sells a product or a service to its customers, it's actually not this kind of hierarchical thing. It actually cuts across that org chart.”

    Listen at 20:09

  21. Ali Ghodsion digital-native companiesPositive21:25

    Digital-native companies are generally further ahead in AI adoption.

    “In general, I think digital natives are probably further ahead.”

    Listen at 21:25

  22. Ali Ghodsion AI inference cost optimizationPositive30:31

    Routing tasks to simpler models and using open source can significantly reduce AI costs.

    “there is a way to spend tokens in a smart way. You can use open source. You can route simpler questions to simpler models. And we actually have a product for this. And you can cut your costs down significantly.”

    Listen at 30:31

  23. Ali Ghodsion DatabricksPositive33:10

    Databricks aims to reduce the number of human blockers in organizational workflows.

    “one of our main goals is reduce number of human blockers.”

    Listen at 33:10

  24. Ali Ghodsion DatabricksPositive38:04

    Ghodsi wants Databricks growth to exceed 100% annually.

    “I'd like to be 100, you know, way over 100.”

    Listen at 38:04

  25. Ali Ghodsion AI-enabled organizational managementPositive38:12

    Superior AI-enabled organizational management currently provides a competitive advantage.

    “Whoever can do this best, you know, at some point, maybe it's all, everyone is using, it's fully diffused and everybody's doing this. But right now, this is big alpha.”

    Listen at 38:12

  26. Going public provides companies greater capital access and marketing visibility.

    “I actually think that you get access to capital. You get much, much better marketing.”

    Listen at 39:38

  27. Ali Ghodsion employee liquidityPositive40:13

    Predictable annual employee liquidity is beneficial.

    “I think the liquidity is a good thing because, you know, people know that they're going to get it every year and it kind of comes and it's like predictable.”

    Listen at 40:13

  28. Ali Ghodsion public-market volatilityNegative42:41

    Public-market volatility can pressure companies into short-term decisions.

    “I don't want the market to force me that, okay, now it's down and now it's, you know, sentiment is terrible this week and now your stock is down.”

    Listen at 42:41

  29. Ali Ghodsion DatabricksPositive43:24

    Databricks will eventually become a public company.

    “We will go public. We are going to be public.”

    Listen at 43:24

  30. Recursive self-improvement requires four specific conditions.

    “So we really need four conditions to be true.”

    Listen at 46:30

  31. Recursive improvement requires training speed to improve superlinearly.

    “it has to be super linear.”

    Listen at 47:21

  32. Recursive self-improvement requires models to use fewer resources over time.

    “we also want it to use less resources”

    Listen at 47:23

  33. Recursive self-improvement requires generalized accuracy improvements across tasks.

    “we're getting better accuracy across the board in a generalized manner.”

    Listen at 47:39

  34. Recursive self-improvement requires continuous repetition of the first three improvements.

    “The fourth condition is that you can actually put this in a loop and all the previous three conditions need to be continuously happening.”

    Listen at 47:58

  35. The recursive self-improvement flywheel does not currently exist.

    “That flywheel doesn't exist right now.”

    Listen at 48:25

  36. Ali Ghodsion recursive self-improvement and AI takeoverNeutral48:52

    Meeting RSI conditions would not guarantee a conscious takeover by AI.

    “even if you satisfy all the four conditions I'm saying, that doesn't guaranteed mean that you'll have... You know, a conscious being that's just, you know, taking over everything”

    Listen at 48:52

  37. Ali Ghodsion AI-enabled cyberattacksNegative50:16

    AI-enabled cybersecurity threats are a real risk.

    “cyber, I do think, is a real thing. And I think it's a real risk.”

    Listen at 50:16

  38. Ali Ghodsion cybersecurity attack-defense asymmetryNegative51:42

    Cyber defense requires preventing every intrusion while attackers need only one success.

    “The asymmetry is that if I'm a defender, I was able to protect my boundaries all the time, except one got through.”

    Listen at 51:42

  39. Humanity will write more code in the next 12 months than ever before.

    “In the next 12 months, we're going to write more code than ever in the history of mankind”

    Listen at 52:28

  40. Ali Ghodsion AI-generated software codePositive52:35

    Newly produced code will be secure because AI systems scan it.

    “So all that code is going to be secure.”

    Listen at 52:35

  41. Ali Ghodsion software vulnerabilitiesPositive53:18

    Code vulnerabilities are finite and will end once the codebase is secured.

    “there's only a finite number of vulnerabilities in the code. Once we've secured that code base, there is no more.”

    Listen at 53:18

  42. Ali Ghodsion AI cybersecurity riskMixed55:58

    Cybersecurity is the biggest current AI risk but is manageable.

    “I think the biggest risk right now is cyber. And I think it's manageable.”

    Listen at 55:58

  43. There is currently no evidence of recursive self-improvement.

    “This RSI, I don't see any evidence of it.”

    Listen at 56:03

  44. Ali Ghodsion AI-assisted medical progressMixed57:37

    AI will accelerate medical progress, but treatments still require human trials.

    “I think progress will be faster, but for a lot of these things, it also has to go through human trials.”

    Listen at 57:37

  45. Ali Ghodsion TEDDYPositive58:28

    TEDDY will accelerate drug discovery substantially.

    “So this will accelerate drug discovery. It's a game changer”

    Listen at 58:28

  46. Crisis Text Line AI detects self-harm risk and helps prevent it.

    “they actually detect. if kids want to do self-harm, and it helps prevent that”

    Listen at 58:41

  47. AI will improve the world overall.

    “And it is going to make the world a better place.”

    Listen at 59:10

  48. Ali Ghodsion enterprise AI adoptionNeutral1:00:10

    Enterprise AI adoption will take approximately 10–15 years.

    “I actually think it's going to take 10 years. I think this is something that's going to take 10, 15 years.”

    Listen at 1:00:10

  49. AI’s effect on GDP will emerge slowly over decades.

    “GDP will move very slowly and it's going to take decades for us to actually get through it.”

    Listen at 1:00:21

Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.

Books & mentions

Some links are affiliate links — PodLume may earn a commission if you buy.

Weekly email

State of the Pod, every Monday

The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.

Listen to the full episode and explore every guest, topic, and moment on PodLume.