Modal is a serverless cloud platform where developers write Python code and Modal runs it on GPUs and CPUs, with products spanning Inference, Sandboxes, Training, Notebooks, Batch , and a broader infrastructure layer aimed at AI workloads. The company was founded by Erik Bernhardsson and Akshat Bubna , and it targets teams building generative AI products, coding agents, model training pipelines, and large-scale inference services.
This launch marks Modal's Series D. The company has crossed $1B in annualized run rate and raised a $750M Series D led by Accel to build new infrastructure for AI applications , at a $15.75 billion post-money valuation, bringing total funding to $1.2 billion . Since the Series C five months earlier, usage has tripled , with Sandboxes, launched in 2023 for running LLM-generated code, now a roughly $300M run-rate product on its own and used for background agents, reinforcement learning rollouts, and vibecoding platforms.
The launch matters because Modal is positioning itself as a default compute layer for the current wave of AI products, with customers including DoorDash's AI platform team, Cognition, Ramp, Linear, and Hebbia building background agents, Runway generating video, Physical Intelligence, Generalist AI, and 1X training robots, and Chai Discovery and Achira running drug discovery . For founders and operators evaluating where to run AI workloads, the Series D signals Modal has the capital and customer base to keep expanding across inference, training, and agent sandboxing as those categories consolidate.
Comments (14)
Sign in to join the discussion.
P
Priya Raghavan18h ago
Love the numbers but curious what gross margin looks like when half your workload is bursty GPU jobs. Annualized run rate is a vibe, net retention is a religion.
R
Reginald Oduya17h ago
A $750M round means the burn plan is spicy. Hope the pricing page keeps up with the GPU bill because the math only works one direction.
T
Tomasz Krawczyk18h ago
The tweet buries the lede under a truncated link. If you're announcing a billion, at least let the sentence finish.
D
Dafne Oyelaran18h ago
15x YoY and the launch graphic looks like it was made at 2am on the back of a napkin. Respect, honestly.
H
Harold Benning17h ago
Is this the thing people use instead of Heroku now? I keep hearing the name but I still deploy with a tarball and a prayer.
K
Kenji Alvarado17h ago
Pinging the API team: can we please get webhook retries with exponential backoff that don't retry on 4xx. I will send cookies.
R
roastbeef17h ago
A $15.75B valuation for making Python functions run on GPUs. My CS professor owes me an apology.
L
Lucia Mendez-Park17h ago
Great raise. Have you considered adding a visual workflow builder, a prompt registry, and a built-in vector DB by Q1? Just stacking the roadmap for you.
K
Kweku A.17h ago
The AI infra market is already consolidating and I think we're one hyperscaler price cut away from every neocloud looking very mortal. Enjoy the moment though.
N
Nadia Boateng17h ago
Fair but 'one hyperscaler price cut' has been the bear case for three years and the neoclouds keep eating. The devs just don't want to touch IAM.
M
Mikhail V.17h ago
Imagine if compute credits were tokenized and devs could stake unused GPU time. Not saying you should. But also.
A
Anya Petridis17h ago
What's the headcount here? If a team under 100 is doing a billion run rate I'm going to go cry into my single-founder dashboard.
S
Sana Qureshi17h ago
Dumb question but if usage tripled since the Series C, why raise $750M now instead of just... running the business? Asking as someone who clearly doesn't know.
B
Bastien Lefèvre17h ago
Fifth AI infra launch I've seen this month and the only one that didn't open with a demo of a chatbot summarizing a PDF. Points for restraint.
Work With Us
Book an intro call to explore our suite of services. We might have a waitlist.
Modal
AI infrastructure that developers love
AI infrastructure that developers love
Tags
TLVC Rating
Community Rating
No ratings yet
to rate this launch.
Modal is a serverless cloud platform where developers write Python code and Modal runs it on GPUs and CPUs, with products spanning Inference, Sandboxes, Training, Notebooks, Batch , and a broader infrastructure layer aimed at AI workloads. The company was founded by Erik Bernhardsson and Akshat Bubna , and it targets teams building generative AI products, coding agents, model training pipelines, and large-scale inference services. This launch marks Modal's Series D. The company has crossed $1B in annualized run rate and raised a $750M Series D led by Accel to build new infrastructure for AI applications , at a $15.75 billion post-money valuation, bringing total funding to $1.2 billion . Since the Series C five months earlier, usage has tripled , with Sandboxes, launched in 2023 for running LLM-generated code, now a roughly $300M run-rate product on its own and used for background agents, reinforcement learning rollouts, and vibecoding platforms. The launch matters because Modal is positioning itself as a default compute layer for the current wave of AI products, with customers including DoorDash's AI platform team, Cognition, Ramp, Linear, and Hebbia building background agents, Runway generating video, Physical Intelligence, Generalist AI, and 1X training robots, and Chai Discovery and Achira running drug discovery . For founders and operators evaluating where to run AI workloads, the Series D signals Modal has the capital and customer base to keep expanding across inference, training, and agent sandboxing as those categories consolidate.
Comments (14)
Love the numbers but curious what gross margin looks like when half your workload is bursty GPU jobs. Annualized run rate is a vibe, net retention is a religion.
A $750M round means the burn plan is spicy. Hope the pricing page keeps up with the GPU bill because the math only works one direction.
The tweet buries the lede under a truncated link. If you're announcing a billion, at least let the sentence finish.
15x YoY and the launch graphic looks like it was made at 2am on the back of a napkin. Respect, honestly.
Is this the thing people use instead of Heroku now? I keep hearing the name but I still deploy with a tarball and a prayer.
Pinging the API team: can we please get webhook retries with exponential backoff that don't retry on 4xx. I will send cookies.
A $15.75B valuation for making Python functions run on GPUs. My CS professor owes me an apology.
Great raise. Have you considered adding a visual workflow builder, a prompt registry, and a built-in vector DB by Q1? Just stacking the roadmap for you.
The AI infra market is already consolidating and I think we're one hyperscaler price cut away from every neocloud looking very mortal. Enjoy the moment though.
Fair but 'one hyperscaler price cut' has been the bear case for three years and the neoclouds keep eating. The devs just don't want to touch IAM.
Imagine if compute credits were tokenized and devs could stake unused GPU time. Not saying you should. But also.
What's the headcount here? If a team under 100 is doing a billion run rate I'm going to go cry into my single-founder dashboard.
Dumb question but if usage tripled since the Series C, why raise $750M now instead of just... running the business? Asking as someone who clearly doesn't know.
Fifth AI infra launch I've seen this month and the only one that didn't open with a demo of a chatbot summarizing a PDF. Points for restraint.