BG2Pod with Brad Gerstner & Bill Gurley

Episode: AI Enterprise - Databricks & Glean | BG2 Guest Interview

Date: December 23, 2025
Guests: Ali Ghodsi (CEO, Databricks), Arvind Krishna (CEO, Glean)
Host: Kral Walker

Episode Overview

In this episode, industry visionaries Ali Ghodsi (Databricks) and Arvind Krishna (Glean) join Kral Walker to dissect the current state of AI in the enterprise. The discussion navigates the reality behind the AI hype, the challenges of enterprise adoption, the promise— and pitfalls—of generative AI, the economics of AI infrastructure, and the evolving software stack. Listeners gain candid insight into what’s truly working, what’s not, and the stakes for businesses and technology leaders as artificial intelligence becomes an operational imperative.

Key Discussion Points & Insights

1. The State of AI: Reality vs Hype

• Super Cycle Context: Kral sets the tone, noting both guests have seen every tech super cycle—from Internet to AI (00:59).
• Enterprise Fog vs Consumer Adoption: While consumer AI has seen mass adoption, enterprise AI projects reportedly fail 95% of the time (01:00).
• Experimentation is Healthy: Arvind reframes high failure rates:

  “If all your projects are failing, that means you’re just not trying enough.”
  (02:13, Arvind Krishna)

2. Use Cases: Where AI is Making a Difference

• Finance (Royal Bank of Canada): AI agents auto-analyze earnings reports, cutting report cycles from hours to 15 minutes (03:32).
• Healthcare (Merck): Transformer models (“Teddy”) enabling next-generation drug discovery (04:41).
• Retail (7-11): Marketing stacks automated to segment audiences and generate materials at scale (05:25).
• Realism: “We’re not just the 5%. We have some of that 95% too,” Ali shares on project success/failure split (06:31).

3. Patterns of AI Success in the Enterprise

• LLMs as Commodities:

  “LLMs have become that way. It doesn’t really matter. This one is better right now, next week that one is better. You can’t even keep up anymore.”
  (06:54, Ali Ghodsi)

• Data is Differentiator: What distinguishes winners is unique, proprietary company data, not the model itself.
• Danger of Demo-Ware: Many cool GenAI demos don’t translate to lasting business value (07:44).

4. Lessons from Failure—Why AI Projects Don’t Work

• Engineering Outpaced by the Speed of Change:

  “You build systems ... and it doesn't seem like a good idea anymore within two weeks because we see a new development.”
  (08:42, Arvind Krishna)

• Change Management: Building and deploying AI in enterprise is still “an engineering art” needing sustained effort and team alignment (03:32, Ali Ghodsi).
• Unexpected Complexity: Even internal ambitions—like auto-prioritizing employees’ weekly goals—remain elusive (09:19, Arvind Krishna).

5. AI vs RPA: Why Today Is Different

• Fundamental Leap:

  “Machines just simply cannot do these kind of things that we saw them do, like writing on their own, having emotion, understanding emotion. It’s fundamental, it’s different ... obvious stuff.”
  (11:31, Arvind Krishna)

• From Rules to Learning: RPA was brittle and rule-based; modern AI learns and adapts (12:32, Ali Ghodsi).
• But: Agents still face the challenge of continuous learning from real-world application—"We haven’t really nailed computer use yet."

6. Guidance for CIOs Planning AI Budgets

• Advice:
  • “Spend more.” (14:37, Arvind Krishna)
  • Favor short-term contracts, easy-to-test products, and broad experimentation due to market volatility (14:39, Arvind Krishna).
• Vendor Caution: Winners are not yet identified; flexibility is key.

7. The AI Infrastructure “Physics Problem”

• CapEx and Value Creation: Kral notes the trillion-dollar investments in compute (Nvidia et al) may outpace realistic revenue (15:51).
• Response:
  • Arvind: Most new AI value will come from eating into the $10T+ services market, not just software (16:43).
  • Ali: “It’s not binary … focus on what’s obviously working, expand that.” Different R&D camps have different time horizons and destinies (17:31).

8. The Three Camps of AI

Ali Ghodsi's framework (17:31):

1. Superintelligence Quest: Chasing AGI via massive scaling (Frontier Labs/Big Tech, high capital burn).
2. Skeptical Scientists: Turing-Award-level researchers, critical of scaling, say "real" AGI is decades away.
3. Practical Builders (themselves): Focused on present economic value;

   “We have AGI by the old standards. Let’s solve actual problems” (20:39). “We just need to expand that 5% to be 10%, 20%, 30%.” (21:38, Ali Ghodsi)

9. Where Will Value Accrue? The Stack Debate

• Commoditization of LLMs: LLM providers will be like “gas stations” (06:54, Ali Ghodsi).
• Data & Governance: Real value lies in proprietary data and in securing/governing it (24:23, Ali Ghodsi).
• Application Layer Dominates: “Most of the value will accrue to the apps … But I just don’t know which apps” (26:10, Ali Ghodsi).

10. The Future of Software Apps

• Not Just Databases: Software is more than a CRUD UI atop a database—workflows, presentations, and product design still matter (28:52, Arvind Krishna).
• Voice/Automation Disruption: Envisioning Zoom as the ultimate data entry point, where meeting info auto-updates CRMs without manual input (31:08, Ali Ghodsi).
• Sprawl and Tools Consolidation: Overabundance of AI note takers is a symptom of rapid change, with consolidation still to come (32:06).

11. Leadership & Adoption in the AI Age

• Real-World AI Agency:
  • Ali uses Databricks’ internal agents to brief him on customer stories, prep presentations, and support go-to-market with the right data—"completely automated" marketing stack (32:44).
  • Arvind relies on a personal “Daily Prep Agent” and AI for company-wide communications, promoting a shift in instincts: “You have to have that belief that AI is a good collaborator ... you’re going to improve the quality of your output.” (36:20)

12. Rapid Fire: Hot Takes on the AI Industry

• Future of Big AI Cos (OpenAI, Anthropic, Gemini): Up—demand and market growth will continue (37:21).

• Is AI in a Bubble?

  “There is an AI bubble ... startups with zero revenue worth $10, $20, $30 billion. That’s a lot.”
  (38:18, Ali Ghodsi) “There are quite a few companies where there’s over-optimism and valuations ... but these companies are going to grow more than non-AI companies.”
  (39:07, Arvind Krishna)

• Long & Short Bets:

  • Ali: Long on agents and speech interfaces; keyboards will disappear. Short on overhyped coding and customer service automation claims (39:57).
  • Arvind: Long on proactive, user-centric AI; not just waiting for users to come, but AI going to them. Big leap will be moving power users to 100% adoption (40:49).

• Favorite AI Tools:

  • Ali: Glean—”I use it all the time.” (41:33)
  • Arvind: AI note takers and deriving corporate knowledge from meetings (41:56)

Notable Quotes & Memorable Moments

• “I think we have AGI. I think we have artificial general intelligence. We really have.” (00:00, Ali Ghodsi)
• “LLMs have become that way... Just compare price.” (06:54, Ali Ghodsi)
• “You build systems and ... it doesn’t seem like a good idea anymore within two weeks.” (08:42, Arvind Krishna)
• “Machines just simply cannot do these kinds of things that we saw them do, like writing on their own, having emotion, understanding emotion... It’s fundamental, it’s different.” (11:31, Arvind Krishna)
• “A lot of the good thing about software companies is that they actually think about how to take that data and present it in a way which drives more productivity from a human.” (28:52, Arvind Krishna)
• “The marketing stack is gonna get disrupted pretty heavily.” (05:25, Ali Ghodsi)
• “If you had [meeting data auto-entering CRMs], that would be the full disruption of the SaaS.” (31:08, Ali Ghodsi)
• “Spend more.” (14:37, Arvind Krishna, on AI budgets)

Timestamps for Important Segments

• Experimentation & Failing is Positive: 02:13–03:08
• Enterprise Use Case Deep-Dive: 03:32–06:39
• Commoditization of LLMs & Data’s Role: 06:54–08:19
• Why AI Projects Fail: 08:42–10:57
• AI vs RPA / Fundamental Differences: 11:22–14:00
• CIO Budgeting & Market Volatility Advice: 14:37–15:51
• AI CapEx “Physics Problem”: 15:51–16:43
• Explaining Three Camps of AI: 17:31–22:40
• Value Accrual Across the Stack: 22:40–28:23
• Software’s Changing Role: 28:52–32:11
• Adopting AI Internally: 32:44–37:21
• Rapid Fire (Future, Bubble, Long/Short): 37:21–42:24
• Glean’s Vision to $1B: 42:47–44:40

Final Vision Statements

• Glean (Arvind Krishna):

  “We want Glean to be this very personal companion for every person in every company in the world ... helping you now with your work, hopefully taking majority of your tasks, automatically working on them before you ask it to.”
  (42:59–44:40)

• Databricks (Ali Ghodsi):
  Focused on scaling real-world, internal automation, and business process transformation—“just focus on solving the actual problems inside organizations ... we already have the AGI we need.” (21:38)

Tone: The conversation is candid, insightful, and accessible, driven by real operator experience—optimism and caution are balanced, with humor and realism interspersed.

For listeners: This episode is a masterclass for enterprise and tech leaders seeking actionable context and a practical compass for investing in and deploying AI today.