Jul 20, 2026 · 43m · sourcery
Hyperscalers Are Out of Capacity? | MongoDB CEO · Sourcery with Molly O'Shea
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 Sourcery, host Molly O'Shea interviews MongoDB CEO CJ Desai at the RAISE conference in Paris to discuss cloud hyperscaler capacity limits, the resurgence of on-premises data centers, and why operational database architecture is the critical foundation powering modern AI agents.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Molly holds 20.5% of the talking time here. How this is scored →
Molly as informed peer 3.0 Guest teaching 3.8 Guest disagreement 0.2 Molly pushing back 0.2
speaking balance: gold is Molly, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 31:55 Reframing open source consensus
CJ gently pushes back against Molly's assertion that the industry is standardizing on open-source models, highlighting enterprise reliance on closed models.
Hardest push from Molly ▶ 31:22 Challenging model standardization narrativeMolly presses on the industry shift towards commoditized open-source models over expensive proprietary options.
Biggest teaching moment ▶ 9:20 Unveiling hyperscaler capacity exhaustionCJ reveals that major cloud providers are rejecting capacity requests from Fortune 100 enterprises, surprising the host with the reality of on-prem revival.
Molly holds their own ▶ 31:22 Synthesizing market trends on model commoditizationMolly demonstrates industry pulse awareness by citing widespread reporting of developers moving down-market to open-source alternatives.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Molly as informed peer | Guest teaching | Guest disagreement | Molly pushing back | Why |
|---|---|---|---|---|---|---|
| OpenAI Stage Discussion & The Data Supercycle | 4 | 4 | 0 | 0 | Molly references recent social media chatter regarding token generation and big data's return. CJ enthusiastically breaks down the three tiers of AI customers adopting MongoDB, explaining the scale-out architecture and unstructured data advantages. | |
| Hyperscaler Capacity Limits & On-Premises Resurgence | 3 | 7 | 1 | 0 | CJ educates Molly on the real-world dynamics of hyperscaler capacity shortages pushing Fortune 100 enterprises back to on-prem data centers. Molly is genuinely surprised by the revelation that large clients are getting capacity refusals. | |
| Real-Time OLTP vs. Analytical Database Architecture | 3 | 5 | 0 | 0 | Molly asks for the core architectural differentiation between MongoDB and analytical warehouses like Snowflake and Databricks. CJ explains the technical distinction between operational OLTP workloads and analytical OLAP queries. | |
| Powering Frontier Labs & Agentic Workloads | 2 | 3 | 0 | 0 | CJ describes how frontier AI labs use MongoDB as a memory layer and real-time inference store. Molly prompts with broad high-level questions while CJ provides technical enterprise context. | |
| Enterprise Agentic Applications & The Data Ecosystem | 3 | 2 | 0 | 0 | Molly asks about adjacent ecosystem players like data labeling startups, and CJ details how MongoDB complements rather than competes with them, followed by his career journey to becoming CEO. | |
| Navigating AI Velocity & Enterprise Architecture Complexity | 3 | 5 | 0 | 0 | CJ illustrates the rapid architectural churn in enterprise AI using an anecdote of a bank's 55-box agentic system. Molly summarizes the takeaway that current enterprise iteration velocity is unprecedented. | |
| Customer Scaling Challenges & Machine-Driven DBAs | 2 | 6 | 0 | 0 | CJ recounts rapid operational scaling challenges for high-throughput clients and explains the historical shift from costly human database administrators to machine-driven automation. | |
| Model Diversity & Deterministic AI Outcomes | 5 | 4 | 1 | 2 | Molly highlights the growing industry trend toward open-source models over closed proprietary systems. CJ offers a nuanced counter-perspective, noting that enterprises maintain heterogeneous model architectures. | |
| Sponsor Segment: MongoDB & AssemblyAI | 2 | 3 | 0 | 0 | Mid-roll ad reads and a speculative discussion on space-based data centers where CJ provides an analytical engineering perspective grounded in solar physics. | |
| Executive Mentors & Talent Density at MongoDB | 4 | 2 | 0 | 0 | Molly probes CJ's executive influences and brings up MongoDB's reputation as a top tier talent incubator. CJ playfully notes mixed feelings about other tech firms poaching his trained staff. | |
| Priorities for the Future & Hot Take | 2 | 1 | 0 | 0 | CJ delivers his core thesis that data infrastructure remains the essential unsung hero of generative AI, concluding the interview on a collaborative and agreeable note. |
Statements from this episode (12)
O'Shea: Cognition AI's Devin Spun Up MongoDB as Its First Task
“The first task that Devin did was spin up MongoDB.”
Desai: Enterprise Agentic AI Is Still in Early Prototyping Stages
“And then you have enterprises who are creating agentic workloads, and that's where we feel we are still early, because I have not seen Agentic applications, whether you look at airlines, right, you look at your banking applications, whichever bank you use, you…”
Desai: Large Paris Enterprise Refuses Public Cloud Over French Sovereignty Rules
“Yesterday I was with a customer, and they said, hey, I just want to tell you, we are not, and it's a large customer based in Paris, we are not going to move our workloads to public cloud. We are going to run it on-prem. Data sovereignty is highest priority. Fr…”
Desai: Hyperscalers are turning away top-50 enterprise customers due to capacity limits
“These hyperscalers, some of them, are running out of capacity, and they're running out of capacity, so this large customer in Texas, speaking to them, they have really good partnership with one of the hyperscalers. They wanted to move more workloads in cloud, …”
Desai: Capacity limits and data privacy are reviving on-prem data centers
“Data centers are back. Nobody is now saying, hey especially in the Fortune 500, and most people get that wrong because everybody thought, hey, all the workloads are just going to move to this hyperscaler. So hyperscaler is running out of capacity and making su…”
Desai: A major US telecom was refused regional cloud capacity
“The other, you know, large telecommunications company in the United States was also similar. They had one particular hyperscaler, and they said this hyperscaler refused to give them additional regional capacity that they needed. So they signed with the second …”
Desai: ElevenLabs runs over 50 million agents on MongoDB
“Out of London, and they have north of fifty million agents, depending on when you look at it, all running on MongoDB.”
Desai: AI startups demand machine-driven database administration over human DBAs
“In database world, when I was at Oracle, the most expensive person you can get is called Oracle DBA, database administrator, and they were the most expensive people that you have to hire when you're running Oracle, and now These AI companies are saying, we don…”
Desai: Enterprises do not standardize on a single AI model
“There is no, like, standardization even when it comes to these models, right? It goes anywhere from open source to closed source, small to large, horizontal to domain specific based on the use cases.”
Desai: Coding agents almost always run on proprietary closed models
“Like, for coding agents, ok, you have Cloud Code, you have Codex, there is a Grok build, many options. Cursor has other options available. So what you find is, in those scenarios, you almost always use See or hear close models, not open source models. Propriet…”
Desai: Space-based data centers are physically possible and could succeed
“So physics-wise, at least from what I have read, It seems like it is, there is a non-zero chance of it succeeding, meaning it will succeed once you put enough attention, resources, focus on it, and like we saw some of the frontier labs, you saw that SpaceX sai…”