👋 Hi, it’s Rohit Malhotra and welcome to the FREE edition of Partner Growth Newsletter, my weekly newsletter doing deep dives into the fastest-growing startups and S1 briefs. Subscribe to join readers who get Partner Growth delivered to their inbox every Wednesday morning.

If you’re new, not yet a subscriber, or just plain missed it, here are some of our recent editions.

He has a newsletter on beehiiv. So do Codie Sanchez, Jay Shetty, David Begnaud, Colin & Samir, and Joanna Stern.

They could publish just about anywhere. They chose beehiiv to build a direct connection with the people who want to hear from them most. And you can do the same, whether you’re world-famous or just at the start of your plan for world domination.

Get started for free.

_{Interested in sponsoring these emails? See our partnership options here.}

Subscribe to the Life Self Mastery podcast, which guides you on getting funding and allowing your business to grow rocketship. 

Previous guests include Guy Kawasaki, Brad Feld, James Clear, Nick Huber, Shu Nyatta and 350+ incredible guests.

Modal raised a $750M Series D at a $15.75B post-money valuation, led by Accel, with Kleiner Perkins joining and existing investors Menlo Ventures, General Catalyst, Redpoint, Lux Capital, Amplify Partners and Bain Capital Ventures participating. The round comes five months after its Series C, as Modal crossed $1B in annualized run rate.

Modal started from one premise: existing infrastructure was not built for AI workloads, given scarce GPUs, slow-loading models and the need to run LLM-generated code in isolation. So it built its own file system, container runtime, scheduler and networking on a capacity pool spanning 20+ clouds. Customers include DoorDash, Cognition, Ramp, Runway and Physical Intelligence.

Its Sandboxes product, launched in 2023 for running AI-generated code, is now a $300M run rate business on its own, with close to 3 billion sandboxes spun up. Modal is also expanding into multi-node training, reinforcement learning, post-training and new inference features.

Modal’s $750M Series D came at a $15.75B post-money valuation, so investors bought less than 5% of the company for three-quarters of a billion dollars. The round lands just five months after the Series C. That gap is short enough that the growth between the two rounds is the story.

Growth drives the round. Modal crossed $1B in annualized run rate, with usage tripling since the Series C and growth above 15x year over year. With 200+ employees, that works out to up to $5M of run rate per person.

Sandboxes carries the argument. Launched in 2023 for running AI-generated code, it is now a $300M run rate business on its own, close to a third of the company. Nearly 3 billion sandboxes have been spun up, and the system scales to a million concurrent executions. TypeSafe AI’s Jev served 1 trillion tokens on Modal in its first three days.

The valuation is the part to sit with. At $15.75B, Modal trades at roughly 15.75x run rate while growing 15x a year. The multiple and the growth rate are almost the same number. Investors are not paying for GPU capacity. They are paying for the layer AI workloads run on.

Accel led the round, with Kleiner Perkins joining as a new name. Every existing investor came back: Menlo Ventures, General Catalyst, Redpoint, Lux Capital, Amplify Partners and Bain Capital Ventures. When an entire cap table invests again five months after the last round, insiders are not hedging. They are buying more of what they already see working.

The numbers make the case. Modal crossed $1B in annualized run rate, with usage tripling since its Series C and growth above 15x year over year. Sandboxes, a 2023 add-on built for running AI-generated code, is now a $300M run rate business on its own, with close to 3 billion sandboxes spun up. The customer list includes DoorDash, Cognition, Ramp, Runway and Physical Intelligence. Those companies span software, robotics and consumer AI, so demand does not hinge on one niche.

The valuation follows from that. $15.75B is roughly 15.75x run rate for a company growing 15x a year. Investors are not paying for hype. They are paying for a growth rate that has not slowed.

The forward bet is the stack. Modal rebuilt the file system, container runtime, scheduler and networking for AI across more than 20 clouds. That is hard to copy, and it is why investors price Modal as a platform rather than a GPU reseller.

That is what $750M actually buys: a push into multi-node training, reinforcement learning and post-training that turns compute customers into platform customers.

One. Timing that fixes itself. AI workloads are outgrowing the cloud they run on. Nearly half of organizations flag GPU and compute costs as a top adoption barrier, and 63.9% of enterprises still deploy AI on hyperscaler-managed platforms built before agents existed. Modal skipped the retrofit. It rebuilt the file system, runtime, scheduler and networking across 20+ clouds, with no capacity commitment. Every GPU shortage and rigid reservation contract makes on-demand compute look less like a convenience and more like the default.

Two. The GPU reseller ceiling, called. Serverless compute was supposed to stay a commodity. Modal turned a 2023 side product into a $300M run rate business. Sandboxes has spun up close to 3 billion isolated environments for agent-generated code, at a time when 55.4% of AI decision-makers name agent reliability as a top production challenge. The ceiling was never real, only untested.

Three. The bet that justifies $750M. Compute is the entry point, not the destination. AWS, Google Cloud and Microsoft hold about 47% of AI infrastructure between them. Modal is moving into multi-node training, reinforcement learning and post-training to win the whole workload lifecycle, not one slice of it. Land the workload, own the stack.

One. AI clouds have the motive and a fresh precedent. In 2025, CoreWeave paid a reported $1.7B for Weights & Biases to put a developer layer on top of its GPU cloud. It now bundles its own Sandboxes product into that platform. The logic was a path from raw compute to the tools developers actually touch. Modal is that same layer with far more surface area: compute, sandboxes, training and inference. The catch is price. At $15.75B, Modal costs roughly nine W&B deals, and every quarter of 15x growth makes the check worse.

Two. Nvidia has the habit. It bought Run:ai for $700M in December 2024 and GPU rental platform Lepton AI for a reported several hundred million in 2025. Nvidia keeps buying the software between its chips and developers. But Modal runs across 20+ clouds, many of them Nvidia’s largest customers. Owning Modal would put Nvidia in competition with the clouds that buy its chips. The likelier path is a deeper partnership.

Three. The real buyer may be the public market. Hyperscalers would pay to win back developer workloads, but antitrust makes that hard. A $1B run rate growing 15x is an IPO story. Strategics buy features. Public markets buy platforms.

One. That the run rate is durable, not spiky. $1B in annualized run rate is a snapshot, not a contract. Usage-based revenue can surge and vanish, like TypeSafe AI’s Jev serving 1 trillion tokens in its first three days. The filing has to show recognized revenue against run rate, net revenue retention, and how much spend is committed versus pay-as-you-go. If customers keep expanding, the platform story holds. If revenue follows launch spikes, it is a burst-capacity business.

Two. That Sandboxes is its own business. At $300M, Sandboxes is close to a third of the company. The S-1 has to break out its revenue and margin. Agent execution priced like software is worth far more than compute resold by the hour.

Three. The unit economics under the GPUs. Modal offers on-demand compute with no capacity commitment while sourcing capacity from 20+ clouds. Gross margin has to show what Modal commits to suppliers, how well it fills that capacity, and whether it can absorb a GPU price swing.

Four. Concentration and dependency. Customers like Cognition and Runway are venture-funded AI companies, and Modal’s suppliers include clouds that sell competing serverless GPU products. Risk factors will show how much revenue rides on both.

Modal did not raise on a story. It raised on usage that tripled in five months, from builders who hit the same wall: scarce GPUs, models slow to load, and agent code that needed somewhere safe to run. That part competitors cannot spin. When AI companies run their workloads on you by default, you are not reselling GPUs. You are becoming the runtime.

The lesson underneath is older than AI. Infrastructure was supposed to belong to the hyperscalers, too capital-heavy and too easy for incumbents to copy. Modal threw out the existing stack, rebuilt it for AI, and found the agent era waiting for a place to execute.

So the $750M is not a bet on compute. It is a bet on who owns the AI workload lifecycle, from sandbox to training to inference, made before AWS, Google and Microsoft repackage serverless GPUs to defend their share. The risk is the same as the thesis: those clouds supply Modal’s capacity and compete for its customers.

Modal has the slope. The open question is whether training, reinforcement learning and post-training turn it into a platform before the hyperscalers turn it into a feature.

Here is my interview with Vineet Budki has spent fifteen years turning volatility into a career - from Emirates cabin crew to CEO of Sigma Capital, a $100M crypto fund with 300+ investments.

If you enjoyed our analysis, we’d very much appreciate you sharing with a friend.

Here are the options I have for us to work together. If any of them are interesting to you - hit me up!

And that’s it from me. See you next week.

What do you think about my bi-weekly Newsletter? Love it | Okay-ish | Stop it