Inception, a Palo Alto-based AI startup, secured $50 million in seed funding to accelerate development of diffusion large language models (dLLMs).
Menlo Ventures led the round, joined by Mayfield, Innovation Endeavors, NVentures (NVIDIA), M12 (Microsoft), Snowflake Ventures, and Databricks Investment. AI luminaries Andrew Ng and Andrej Karpathy participated as angel investors.
What Makes Inception Different?
Inception is reimagining how AI models generate text, code, and voice outputs. Unlike traditional autoregressive models like GPT that predict text word by word, Inception's approach generates entire responses through iterative refinement.
The company launched Mercury, the first commercial-scale diffusion language model, in February 2025. Mercury achieves speeds exceeding 1,000 tokens per second—up to 10x faster than models from OpenAI, Anthropic, and Google while maintaining comparable accuracy.
The Brains Behind the Breakthrough
CEO Stefano Ermon co-invented the diffusion methods powering Midjourney and OpenAI's Sora. He founded Inception in 2024 alongside Aditya Grover and Volodymyr Kuleshov, professors from Stanford, UCLA, and Cornell.
The founding team pioneered Flash Attention, Decision Transformers, and Direct Preference Optimization. Their engineering roster includes veterans from DeepMind, Microsoft, Meta, OpenAI, and HashiCorp.
Why This Funding Matters Now?
AI inference costs are becoming the biggest barrier to scaling deployments. Ermon states that while training gets faster, “inefficient inference is becoming the primary barrier and cost driver to deployment”.
Diffusion models process operations simultaneously rather than sequentially, dramatically reducing latency and compute costs. This makes them perfect for real-time applications like voice agents, live code generation, and interactive interfaces.
The fresh capital will fund product expansion, team growth, and deeper research into text, voice, and coding applications.
Investment Breakdown
| Parameter | Details |
|---|---|
| Funding Amount | $50 million |
| Round Type | Seed |
| Lead Investor | Menlo Ventures |
| Co-Investors | Mayfield, Innovation Endeavors, NVentures, M12, Snowflake Ventures, Databricks Investment |
| Angel Investors | Andrew Ng, Andrej Karpathy |
| Valuation | Undisclosed |
| Previous Funding | $6 million (seed) |
| Total Raised | $56 million |
Key Questions About This Round
Who led this investment?
Menlo Ventures led the $50 million seed round, with participation from major tech venture arms including NVIDIA, Microsoft, Snowflake, and Databricks.
What will the funds be used for?
Inception will accelerate product development, expand research and engineering teams, and advance diffusion systems for real-time text, voice, and coding applications.
What makes Mercury different from ChatGPT?
Mercury uses diffusion technology that generates text through parallel processing rather than sequential prediction, achieving 5-10x faster speeds while maintaining similar quality.
What's Next for Inception?
Mercury has already been integrated into development tools including ProxyAI, Buildglare, and Kilo Code. The model is also available through Amazon Bedrock Marketplace and Amazon SageMaker JumpStart.
Inception faces competition from established players like OpenAI and Anthropic, but their radically different approach to inference could reshape AI economics. The company's next challenge is proving diffusion models can scale beyond specialized use cases to general-purpose applications.
With backing from top-tier investors and proven technical leadership, Inception is positioned to challenge the autoregressive model monopoly that has dominated AI development.
Company Information
| Platform | Link |
|---|---|
| Website | inceptionlabs.ai |
| linkedin.com/company/inception-labs-ai | |
| X (Twitter) | @inception_ai |
| Contact | [email protected] |
| Location | Palo Alto, California |
Sources
BONUS: Get our $200 “AI Mastery Toolkit” FREE when you sign up!
