Dec 23, 2025 · 45m · bg2-pod
AI Enterprise - Databricks & Glean | BG2 Guest Interview · Bg2 Pod
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of BG², host Apoorv Agrawal interviews Databricks CEO Ali Ghodsi and Glean CEO Arvind Jain to examine the real-world state of enterprise AI adoption, addressing failure rate myths, ROI calculation, foundation model commoditization, proprietary data strategy, and internal executive AI workflows.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
Brad and Bill as informed peer 4.1 Guest teaching 5.5 Guest disagreement 3.0 Brad and Bill pushing back 2.0
speaking balance: gold is Brad and Bill, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 20:10 Ali dismisses future AGI debates as false premises
Ali directly rejects the standard industry narrative about pursuing AGI, forcefully arguing that by historical definitions dating back to 2009, AGI is already solved.
Hardest push from Brad and Bill ▶ 15:53 Host pushes on the physics of AI Capex payoffThe host challenges the guests on macroeconomic fundamentals, laying out hard hardware spend numbers and questioning how a trillion dollars of revenue can materialize against a $400B software baseline.
Biggest teaching moment ▶ 17:31 Ali outlines the three distinct camps of AI developmentAli educates the host with an analytical taxonomy separating superintelligence scaling labs, classical Turing-award researchers, and practical enterprise builders.
Brad and Bill hold their own ▶ 15:53 Host breaks down the semiconductor and software revenue mathThe host commands the dialogue by citing precise capex figures, hardware depreciation realities, and total addressable software revenue to frame the economic tension.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Brad and Bill as informed peer | Guest teaching | Guest disagreement | Brad and Bill pushing back | Why |
|---|---|---|---|---|---|---|
| Enterprise AI Adoption and the 95% Failure Rate Myth | 4 | 5 | 3 | 2 | The host sets up the dilemma between consumer adoption and the MIT report citing a 95% enterprise failure rate. Arvind reframes the statistic positively, noting that a high failure rate in early experimentation is expected and desirable. | |
| Production Enterprise AI Case Studies Across Industries | 3 | 6 | 2 | 1 | Ali delivers detailed enterprise case studies across finance, healthcare, and retail to illustrate where AI generates concrete value. The host primarily listens as Ali contrasts real production pipelines with superficial AI demos. | |
| Proprietary Data Strategy vs. Commodity LLMs | 5 | 5 | 4 | 2 | Ali assertively labels foundation models as commodities interchangeable like gasoline, placing all strategic value on proprietary enterprise data. The host aligns with this view, reinforcing Altimeter's data strategy framework. | |
| Internal AI Experiments and Engineering Failures | 3 | 5 | 1 | 1 | Arvind candidly walks through Glean's internal development hurdles, including abandoned fine-tuning efforts and the difficulty of building automated executive rollup tools. The dynamic is reflective and collaborative. | |
| Generative AI Agents vs. Robotic Process Automation | 5 | 6 | 4 | 2 | The host asks if generative AI is repeating the fizzled trajectory of RPA. Both guests firmly reject the comparison, with Ali breaking down the architectural difference between brittle rule-based automation and probabilistic learning systems. | |
| Practical AI Budgeting and Vendor Strategy for CIOs | 7 | 5 | 3 | 5 | The host presents detailed Capex math, noting half a trillion in hardware spend requires a trillion in AI revenue against a $400B software baseline. Arvind responds by explaining that AI spend captures the much larger services industry market rather than traditional software budgets alone. | |
| The Three Camps of AI and Defining Current AGI | 3 | 8 | 5 | 1 | Ali delivers a structured breakdown of the AI landscape into three camps, forcefully arguing that the tech industry already achieved AGI under its historical definition and is now moving goalposts unnecessarily. | |
| Enterprise Value Accrual: Models, Data, and Applications | 4 | 5 | 2 | 2 | The discussion turns to enterprise value capture across data, intelligence, and application layers. Ali draws parallels to the 1998–2000 internet cycle to explain why value will inevitably gravitate to applications and governance. | |
| The Future of SaaS Applications and Automated Data Entry | 4 | 6 | 4 | 3 | The host asks whether standard SaaS applications will be reduced to commodity databases. Arvind dismisses this as an oversimplification, while Ali highlights automated multimodal capture as the true vector for SaaS disruption. | |
| Executive Workflows and Internal AI Adoption at Scale | 3 | 4 | 1 | 1 | Ali and Arvind outline their personal daily AI usage and internal corporate automation strategies, detailing how change management rather than model capability is the primary bottleneck. | |
| Rapid-Fire AI Market Outlook, Long/Short Bets, and Favorite Tools | 4 | 5 | 4 | 2 | In the rapid-fire section, Ali bluntly affirms the existence of an AI bubble in early-stage pre-revenue startups and superintelligence research, while identifying voice interaction as high-upside and coding automation as overhyped. |
Statements from this episode (24)
Jain: Most enterprise employees use ChatGPT and Claude daily
“Everybody in your company is probably using ChatGBD and Claude and other tools on a daily basis.”
Jain: 95% enterprise AI failure rate is desirable for tech experimentation
“The thing that I feel you know, is happening in enterprises, you hear these 95% of projects fail, But, like, you know, like, that's actually what you want. Like, you, like, when you are actually experimenting with new technology, if all of your projects are fa…”
Ghodsi: RBC AI agents produce equity research reports in 15 minutes
“Royal Bank of Canada built agents with us that basically take, As soon as an earnings report comes out. So equity research analyst, their job is to put together these, you know, reports that say, like, you know, this is a buy, this is a, you know, hold, and so…”
Ghodsi: Merck developed transformer model TEDDY for AI drug discovery
“In healthcare, we have you know, customer Merck that in the life science space created a model called TEDDY. TEDDY stands for Transformer Enabled. Drug discovery. And this is a transformer model, kind of just like large language models that can predict the nex…”
Ghodsi: Large language models have become interchangeable commodities
“Yeah, look, I think the LLM is a commodity. People are not saying that, but it is a commodity. Like, and, you know, when I took econ classes, commodity was when it's interchangeable. Like, you can get gas from this gas station, or you can get gas from that gas…”
Ghodsi: Enterprise AI advantage lies in proprietary data and processes
“It really comes down to your company. What data does your company have that's special? That your competitors don't have. Can you leverage that? And can you build AI that really understands that data? Because that's not a commodity. There's not an AI out there …”
Jain: Glean abandoned internal model fine-tuning for pre-built models
“Some of our fine tuning work, building, building models for a specific use case within our product, like, you know, didn't really pan out for us. And ultimately the choice was that, you know, we can go with already built models, whether they are small open sou…”
Jain: Generative AI will not fizzle out like robotic process automation
“It's fundamental, it's different, and the, and that's why, like, you know, I don't think, you know, we, this technology is going to fizzle out. And it's not like, you know, you don't have to be, like, a financial expert or, you know, like, sort of a deep think…”
Ghodsi: High-profile startups attempting to replace RPA with GenAI have failed
“There has been many startups that have failed in the generative AI. We're going to replace RPA with generative AI models. There's many startups that failed actually that I know of, like pretty some high profile ones.”
Ghodsi: Desktop AI automation stalls because frozen models lack continuous learning
“The biggest problem is that you bake a model, and that's where it's learned everything it needs to learn, and then you freeze it, and then you launch it, and then maybe you give it some context, but that's it. It's frozen. So therein lies the problem that, you…”
Jain: CIOs should sign shorter AI contracts as winners remain unclear
“What we tell people is that I think the winners are yet to be identified and so experiment with more vendors, do shorter term contracts.”
Agrawal: Global software industry generates roughly $400 billion in annual revenue
“This, and just to put this in context, the entirety of the software industry earns about four hundred billion dollars of revenue.”
Jain: AI will capture massive revenue from the 25x larger services market
“Going to grab a lot of revenue that actually today is in services industry, which is 25 times larger than software industry. So this is, there's a lot of spend that is going to move. I mean, the spend that you see happen on AI is actually sort of, you know, th…”
Ghodsi: We already have AGI based on historical definitions
“I think we have AGI. I think we have artificial general intelligence. We really have it. We absolutely have it. It's like, Anyone who says we need to get to AGI, that's like, it's a, it's false premise to start with. We already have AGI.”
Ghodsi: Enterprises do not need superintelligence to capture massive AI value
“We have the AGI we need. Let us just focus on solving the actual problems inside the organizations, and I think we can already, that's enough to automate a lot of the tasks and get huge economic value out of it. We don't actually need superintelligence for tha…”
Jain: AI intelligence layer may capture half of enterprise value
“I know we do think that the intelligence layer is actually going to be a pretty thick one. Maybe, you know, it's, it will capture half of the enterprise value.”
Ghodsi: Most AI economic value will accrue to applications
“But I do think most of the value will accrue to the apps. So it's kind of, and I think that's common sense. It's just, I just don't know which apps.”
Jain: SaaS leaders will not be relegated to just backend databases
“So I think ultimately, like, software is an end-to-end stack in my opinion, and all of these companies, you know, I don't think they're going away. I don't think, you know, they're going to relegate it to becoming a database.”
Ghodsi: Users will speak rather than type to software within two years
“That I think is going to happen, yeah, that will happen in the next couple of years, and even Glean, you'll, you won't be wanting to type, you'll want to talk to it.”
Jain: Glean records every internal and external meeting when permitted
“In, in Glean, we have this policy where we record every single meeting, internal meeting, external meeting if our customers allow because there's so much information, you know, in there.”
Ghodsi: Databricks has 6,000 GTM staff and 3,000-4,000 R&D employees
“Databricks is a big, you know, 6000 person go to market org and three, 4000 person R&D org, and then there's some back office stuff.”
Ghodsi: Speech interfaces will completely eliminate physical keyboards
“Like I think keyboards are kind of basically going to disappear. Completely. We haven't actually nailed speech. I know it feels like we have, but we haven't, because you're still using your keyboard. So as long as you're using a keyboard, we haven't nailed spe…”