BREAKING: Inside AI's… | Sourcery Newsletter Summary

Sourcery Newsletter | Molly O'Shea | MongoDB CEO CJ Desai

1. Key Themes

The Data Layer Is the Durable Downstream Winner of the AI Cycle

Every AI application — regardless of which model it uses — generates data that must be stored and served. This makes the database layer a model-agnostic beneficiary of the entire AI wave.

"Data is the unsung hero and data is back. You cannot create an AI application without a great data layer, and your AI application is as good as your data." "Regardless of the era.. i.e. internet era, mobile era, after iPhone, now the AI era.. you always need a data layer."

Hyperscaler Capacity Constraints Are Forcing a Structural On-Prem Comeback

The conventional wisdom that all workloads would migrate to public cloud is cracking under real supply constraints — even for the largest enterprise customers.

"You see these hyperscalers, some of them are running out of capacity… the hyperscaler said, 'Sorry, we don't have a capacity.' And they are one of the top 50 customers for that hyperscaler." "Most people get that wrong because everybody thought all the workloads are just gonna move to these hyperscalers."

Data Sovereignty Is Becoming a First-Order Infrastructure Priority

Regulatory and geopolitical forces are pushing large enterprises — especially in Europe — to keep data off public cloud entirely, creating a structural tailwind for multi-cloud and on-prem database solutions.

"Data sovereignty is highest priority. French regulations forces us to have that as the highest priority." "They chose MongoDB because we work in multiple clouds. We are not just a database that works in only one cloud, and they really like that for resiliency purposes."

Enterprise AI Architecture Is Exploding in Complexity — and Still Early

Real-world agentic deployments are far more complicated than demos suggest, with dozens of new tooling categories emerging simultaneously. Most enterprise use cases are still in prototyping, not production.

"I wanna say 55 different boxes for the agentic architecture." [describing a major NY bank's setup] "Half of them… did not exist a year earlier, things like observability, security, evaluations, guardrails." "I have not seen customer-facing agents running at scale in airline or banking apps yet, and most of what he sees is prototyping."

AI Spending Is Scaling Rapidly, With Value Spreading Across the Stack

Enterprise AI spend has grown from $1.7B in 2023 to $37B in 2025 — and the value is distributing more broadly than in prior software eras, with the data layer sitting alongside (not below) the model.

"Menlo Ventures' enterprise survey puts AI spending at $37 billion in 2025, up from $1.7 billion in 2023, about 6% of the global software market." "The recurring finding is that the top reason agents break in production is bad data, not the model or the framework."

2. Contrarian Perspectives

On-Prem Is Not Dead — It's Returning for Non-Obvious Reasons

The market consensus has been that cloud migration is a one-way, irreversible trend. Desai's customer evidence directly contradicts this: a Fortune 100 Texas company kept a data center it was planning to shut down because its hyperscaler refused capacity, and a US telecom signed with a second cloud provider just to get regional access.

"Most people get that wrong because everybody thought all the workloads are just gonna move to these hyperscalers… Between the capacity crunch and the pull to keep proprietary data in-house, he sees on-prem and private data centers coming back, and not only at regulated companies."

There Is No Enterprise AI Model Standardization — And There Won't Be Soon

The consensus narrative is that enterprises will consolidate on one or two preferred models (similar to cloud provider consolidation). Desai expected this too — and was wrong.

"I had a very simple viewpoint of the world from a customer perspective. Oh, are you standardizing on this particular model." Instead: "There is no standardization even when it comes to these models. It goes anywhere from open source to closed source, small to large, horizontal to domain specific based on use cases." "I have not seen a complete pivot towards open source or a complete pivot towards proprietary."

The "If You Build It, They Will Come" Era of Infrastructure Is Over

The classic infrastructure playbook — build ahead of demand and let adoption follow — has been invalidated by the pace of AI. Companies are being caught off guard by demand surges they didn't anticipate.

"You can't come and think of, if I build it, they will come. That era is gone." "There is no solution. You constantly have to pivot or change the roadmap… This is the slowest. I think it is going to get even faster."

3. Companies Identified

MongoDB (NASDAQ: MDB) Document database company; ~$25B market cap; 65,200+ customers; ~75% of Fortune 100 Central subject of the article — positioned as the operational database layer for AI applications and agentic workloads

"Frontier labs, I would consider the holy grail for us, with growth he calls faster than linear."

ElevenLabs AI voice/audio company; AI-native startup Scale case study demonstrating MongoDB's fitness for agentic workloads

"They have north of 50 million agents, depending on when you look at it, all running on MongoDB. That also gives us a lot of confidence that we have the right architecture for agentic workloads."

Databricks Data and AI platform; still private; ~$188B valuation (Coatue-led round, July 2026); ~$5.4B ARR, 65%+ growth; positive FCF Competitor/market context; pushing into operational database territory with Lakebase (a Postgres database aimed at AI agents)

"Databricks pushes Lakebase, a Postgres database aimed at AI agents, into operational territory Oracle has long held."

Snowflake (NYSE: SNOW) Cloud data platform; ~$93B market cap; ~$5B trailing revenue; guiding $5.84B product revenue for FY2027 (~31% growth) Competitor in the analytics/OLAP space; named alongside MongoDB as a Citi top software pick

"In July 2026, Citi named MongoDB, Snowflake, and Palantir its top three software picks for the year."

Palantir (NYSE: PLTR) AI/data analytics platform for enterprise and government Named as one of Citi's top three software picks alongside MongoDB and Snowflake

"In July 2026, Citi named MongoDB, Snowflake, and Palantir its top three software picks for the year."

Zomato Food delivery company; India-based Enterprise customer example running on MongoDB Atlas

"The customers run across travel, entertainment, and media, and include the food-delivery company Zomato, on MongoDB Atlas."

Emergent / Base44 Vibe-coding / AI-native development platforms Examples of AI-native companies built on MongoDB

"Vibe-coding platforms like Emergent and Base44 are built on MongoDB."

Brex Intelligent finance platform (cards, expenses, travel, bill pay, banking) Sponsor; notable customer list includes OpenAI, Anthropic, Vercel

"Trusted by OpenAI, Anthropic, Vercel, Granola, Deepgram, & Sourcery."

AssemblyAI Speech-to-text and voice AI API platform Sponsor; used by Granola, ClickUp, HeyGen

"Millions of developers use AssemblyAI to power their voice AI apps & features."

Hugging Face Open-source AI model repository Referenced as source for domain-specific open-source models used by enterprises

"Some pull domain-specific open-source models off Hugging Face."

4. People Identified

CJ Desai CEO of MongoDB (since November 2025); former COO of ServiceNow (scaled $1.5B → $10B+ revenue); former head of product & engineering at Cloudflare; started career at Oracle Primary interview subject; brings direct customer intelligence from 10–12 customer conversations per week

"Data is the unsung hero and data is back. You cannot create an AI application without a great data layer, and your AI application is as good as your data."

Dev Ittycheria Former CEO of MongoDB; 11-year tenure prior to Desai Context for the leadership transition and the mandate Desai inherited

"CJ took over MongoDB… after Dev Ittycheria's 11-year run, with a mandate to reposition the company as the default modern database for AI applications."

Mati (founder of ElevenLabs) Founder of ElevenLabs (AI voice platform) Referenced for validating MongoDB's talent quality — specifically citing MongoDB salespeople and engineers as top recruiting targets

"Everybody told me we should hire salespeople from MongoDB because you guys do a really good job of training them. Same thing with our engineers because they understand distributed systems & data really well."

5. Operating Insights

Auto-Scaling and Autonomous Database Administration Are Now Table Stakes for AI Customers

Fast-moving AI companies don't staff database administrators. They expect the database to manage itself. Operators building infrastructure products for AI-native companies must design for zero-ops, machine-managed administration — or they'll lose deals.

"Why am I calling you and telling you that I'm gonna run out of capacity? Why couldn't you see it?… We don't have time to hire these people. If we are using your software, it should have its own machine-based database administration."

For Enterprise AI, Multi-Cloud Compatibility Is a Sales Requirement, Not a Nice-to-Have

As enterprises end up with workloads split across GCP, Azure, and AWS — sometimes because one provider couldn't accommodate them — infrastructure vendors that are locked to a single cloud are at a structural disadvantage.

"They chose MongoDB because we work in multiple clouds. We are not just a database that works in only one cloud, and they really like that for resiliency purposes."

Stay in Constant Customer Proximity — Especially at the CEO Level

Desai's operating rhythm of 10–12 customer conversations per week gives him real-time intelligence on frontier trends before they surface in analyst reports. For CEOs of infrastructure companies, this cadence is a competitive moat.

"CJ runs on a cadence of speaking to 10 to 12 customers a week, and in this conversation he brings that view straight from frontier labs, AI-native startups, and Global 2000 enterprises."

6. Overlooked Insights

Memory Is Becoming a Discrete Infrastructure Layer — Not Just a Feature

The article briefly references dedicated memory (storage carrying context across sessions) becoming standard infrastructure in 2026 the same way vector databases did in 2024. This suggests an emerging product category, not just a database feature, that is still relatively undercovered in investment discourse.

"Research from groups like mem0 sees dedicated memory, storage that carries context across sessions, turning into standard infrastructure in 2026 the way vector databases did in 2024."

Robotics and Physical AI Are a Stealth Demand Driver for Operational Databases

While most AI database discussions focus on software agents and LLM applications, physical AI (robots generating continuous sensor and action data) is quietly emerging as a significant and distinct workload category — one that Desai is already seeing in production.

"Every robot is generating data as they sense thing, as they act on things, and then all that data gets stored in MongoDB."