← STARTUPSAI INFRASTRUCTURE · BASELINE 2026-04-20

Modal

confidence: high · 3 watch questions · 7 evidence citations

01 TEAM & SIZE

Co-founders Erik Bernhardsson (CEO, ex-Spotify music recs, ex-Better CTO) and Akshat Bubna (CTO). Founded 2021, HQ NYC. Headcount ~132 as of Mar 2026 per third-party tracker (unverified — Modal does not publish official count; LinkedIn shows "51-200"). Bernhardsson remains the public face via high-traffic engineering blog (erikbern.com) and active X presence.

02 FUNDING & STAGE

Total raised ~$111M. Series A $16M Oct 2023 (Redpoint led, a16z participated). Series B $87M announced Sep 29 2025, led by Lux Capital at $1.1B post. Per TechCrunch Feb 11 2026, in "early" talks for new round at ~$2.5B valuation (General Catalyst rumored lead) with ~$50M ARR cited — Bernhardsson publicly downplayed as "general conversations." Stage: late Series B / pre-C.

03 PRODUCT STATE

Serverless Python-native platform for AI workloads: inference, training, batch, sandboxes, notebooks. Key wedges: sub-second GPU cold starts, function-level autoscaling, local-feeling DX (decorator API). Free tier $30/mo compute. Competes with RunPod, Baseten, Replicate, Fireworks on serverless-GPU; against Anyscale/Coreweave upstream. Product breadth expanded 2025 into Sandboxes (agent code-exec) — Meta reportedly used for "Code World Models" spinning up thousands of concurrent sandboxes.

04 GTM MOTION

Bottom-up PLG: free-tier → self-serve → enterprise. Channels: Bernhardsson's blog + X top-of-funnel, OSS modal-examples, case studies, Latent Space / MLOps podcasts. Named customers span foundation labs (Mistral, Allen AI), AI-native apps (Suno, Lovable, Cognition, Harvey, Decagon, Cartesia, Codegen, Phonic, Chai), and enterprises migrating ML (Substack, Ramp, Quora, Scale, Meta project use). "10,000+ teams" cited (unverified).

05 CORE THESIS

Bet: the serverless-GPU layer becomes the default abstraction for AI-native developer teams who refuse to manage K8s/GPU fleets. Modal wins on DX + cold-start perf while hyperscalers remain awkward for bursty inference. Risk: commoditization pressure from Baseten/Fireworks, hyperscaler price cuts, and AWS/GCP native serverless-GPU maturing. Bear case (Nextword Substack): AI-infra startups face margin squeeze as GPU supply normalizes.

06 PUBLIC SIGNALS INVENTORY

Blogmodal.com/blog (Engineering, Customer Stories, Tutorials, News categories) — RSS feed URL not surfaced on index; likely /blog/feed.xml (unverified)
RSS
GitHubgithub.com/modal-labs — 78 repos, pinned: modal-examples (1.2k), modal-client (466), quillman (1.2k), turbo-art, vprox
Twitter@modal_labs (corporate), @bernhardsson (CEO, very active)
LinkedInlinkedin.com/company/modal-labs
Careersmodal.com/careers
YouTubeModal Labs channel (tutorials, launches)
PressLatent Space, Software Engineering Daily (Jul 2025), TechCrunch, SiliconAngle; Contrary Research deep-dive

07 52-WEEK QUESTIONS

  1. Q1
    Does the rumored $2.5B round close, and at what ARR multiple — will disclosed ARR exceed $50M by Q3 2026?
  2. Q2
    Do Mistral, Meta, or a frontier lab convert from project-use to anchor customer / displace in-house infra?
  3. Q3
    Can Modal defend GPU gross margin as Baseten/Fireworks price-war and AWS Bedrock serverless-GPU matures — headcount/burn trajectory signals?

08 WEEKLY TIMELINE

0 signals

TIMELINE EMPTY · WEEK 0

Auto-feed fetcher ships Month 2. Manual diffs via Friday script start Week 2.

09 EVIDENCE

7 items · expand
  1. S-5-001
    Modal Series B post ($87M, $1.1B, Lux-led, Sep 29 2025)
  2. S-5-002
    TechCrunch Feb 11 2026: $2.5B talks, ~$50M ARR
  3. S-5-003
    modal.com/customers — 24 named logos incl. Suno/Substack/Ramp/Lovable/Quora/Decagon
  4. S-5-004
    Contrary Research business breakdown
  5. S-5-005
    github.com/modal-labs — OSS footprint
  6. S-5-006
    SiliconAngle Series B coverage
  7. S-5-007
    Software Engineering Daily podcast (Bernhardsson, Jul 2025)

10 CONFIDENCE & UNKNOWNS

expand

High confidence on funding history, product surface, named customers, founder identity. Medium on headcount (132 is tracker-derived, LinkedIn band "51-200"). Unverified: exact ARR ($50M is single-sourced leak), "10,000+ teams" claim, RSS URL, Quizlet as customer (not found on customers page — drop from hints). Gross margin and net retention unknown.

11 RELATED