Magic

Magic
Magic
  • Headquarters

    San Francisco, CA
  • Founded

    2022
  • CEO / Founder

    Eric Steinberger
  • Employees

    11-50
  • Funding

    Series D+ — $515M
  • Products

    1 product listed
  • Avg. AI Score

    6.0/10
  • Website

    magic.dev/

Profile updated

Products
1
Avg AI Score
6.0/10
Founded
2022
Funding
$515M
About Magic

Magic AI, Inc. is an artificial intelligence company founded in 2022 by Eric Steinberger and Sebastian De Ro, headquartered in San Francisco. The company develops frontier-scale code models with the stated goal of building safe AGI by automating both AI research and software engineering at scale.

Magic's flagship research focuses on ultra-long context windows — demonstrated at 100 million tokens — enabling models to reason across entire codebases simultaneously. Its technical stack combines domain-specific reinforcement learning, inference-time computation, and custom supercomputing infrastructure including thousands of NVIDIA GB200 GPUs. The LTM-1 model, released in 2024, established the company's position in long-context code modeling.

Magic raised a $23 million Series A in February 2023, followed by a $117 million Series B in February 2024 led by Nat Friedman, Daniel Gross, CapitalG, and Elad Gil. In August 2024 the company closed a $320 million round backed by Eric Schmidt, Atlassian, Sequoia, and Jane Street, bringing total funding to approximately $515 million as of 2026. The company operates with a team of roughly two dozen engineers and researchers.

Specialties
AI code generationlarge language modelssoftware engineering automationreinforcement learninglong context modelsAGI researchpre-traininginference optimization
Financials

Dated, source-cited figures. Estimates are marked.

Valuationest.
$1.5B
siliconangle.com
Total Raised
$515M
magic.dev
Valuation over time
$500M(Feb 20, 2024, est.)$1.5B(Jul 2, 2024, est.)
Total Raised over time
$23M(Feb 6, 2023, est.)$145M(Feb 20, 2024, est.)$465M(Aug 29, 2024, est.)$515M(August 2026)
Products by Magic

All products by Magic reviewed by our AI panel of experts.

How Our Review Panel Scores Magic

Six independent reviewer personalities assessed Magic's product on their own terms. They disagreed by 2.0 points — The Domain Practitioner was the hardest to convince, The Domain Strategist the most positive.

The Domain Strategist
7.2
The Decision Maker
6.2
The Power User
6.2
The Skeptic
6.2
The Finance Lead
5.2
The Domain Practitioner
5.2

Averaged across 1 reviewed product in our catalog. Scores are out of 10.

Competitive Landscape

Who Magic competes with, and how they differ.

Devin (Cognition) logo
Devin (Cognition)7.8/10

Cognition's Devin is also pitched as an autonomous "AI software engineer" rather than an in-editor assistant, the closest positioning match to Magic; Cognition was in funding talks at a $25-40B valuation in 2026, far outscaling Magic's ~$500M raised to date.

1 product reviewed on TopReviewed
Cursor (Anysphere) logo
Cursor (Anysphere)7.8/10

Cursor is an in-IDE AI pair-programmer with mass developer adoption rather than a fully autonomous agent; SpaceX acquired Anysphere for a reported $60B in June 2026, giving it vastly greater resources than Magic.

1 product reviewed on TopReviewed
Augment Code logo
Augment Code7.9/10

Augment Code targets large enterprise codebases with long-context retrieval-augmented completion, the closest technical overlap with Magic's long-context bet, but ships as an IDE plugin rather than a standalone autonomous agent.

1 product reviewed on TopReviewed
Windsurf logo
Windsurf7.8/10

Windsurf (formerly Codeium) built an agentic coding IDE and raised in the same 2024 funding wave as Magic; after OpenAI's proposed $3B acquisition collapsed in 2025, Google licensed its technology and hired key staff instead, leaving Windsurf's independent roadmap uncertain.

1 product reviewed on TopReviewed
Anthropic (Claude Code) logo
Anthropic (Claude Code)8.2/10

Anthropic's Claude Code is an agentic CLI coding tool built on its general-purpose frontier models rather than a code-specialized architecture, competing for the same "autonomous engineering" use case but backed by vastly greater model-scale funding.

3 products reviewed on TopReviewed
P
Poolside AI

Poolside is Magic's closest peer by business model — a foundation-model lab training code-specific models rather than an application layered on others' models — and struck a $6B Nvidia licensing deal alongside a $1B raise at a $12B valuation in August 2026, well ahead of Magic's scale.

Company Timeline

Key milestones in Magic's history, with sources.

  1. Funding

    $320M round — Eric Schmidt, Atlassian, Sequoia, CapitalG, Jane Street

    Magic raised $320 million from investors including former Google CEO Eric Schmidt, Atlassian, Sequoia, CapitalG, Jane Street, Nat Friedman and Daniel Gross, bringing total funding to date to nearly $465 million.

    techcrunch.com
  2. Partnership

    Google Cloud partnership for Magic-G4 and Magic-G5 AI supercomputers

    Magic partnered with Google Cloud to build two AI training clusters, Magic-G4 (Nvidia H100) and the planned Magic-G5 (Nvidia Blackwell), aiming for a combined 160 exaflops of compute.

    techcrunch.com
  3. Launch

    LTM-2-mini model — 100M-token context window

    Magic unveiled LTM-2-mini, a custom architecture with a 100-million-token context window (about 10 million lines of code), which TechCrunch reported as the largest context window of any commercial model at the time.

    techcrunch.com
  4. Funding

    Series B — $117M led by NFDG Ventures

    Magic raised a $117 million Series B led by NFDG Ventures (Nat Friedman and Daniel Gross), with CapitalG and Elad Gil also participating, bringing total funding to around $145 million.

    voicebot.ai
  5. Funding

    Series A — $23M led by CapitalG

    Magic raised a $23 million Series A led by Alphabet's CapitalG, with participation from Elad Gil, Nat Friedman, and Amplify Partners, for its AI coding-assistant platform.

    siliconangle.com
  6. Founded

    Magic founded by Eric Steinberger

    Magic was founded in 2022 by Eric Steinberger (CEO) to build an AI system for automating software engineering.

    pulse2.com

Explore AI Software Reviews

Browse multi-perspective AI panel reviews across hundreds of AI tools, agents, and platforms. Find the right software with insights from CTO, Developer, Marketer, Finance, and User perspectives.