Language

Synthszr Charts — die großen AI-Marken im Wettkampf ums Podium

ML Intern

#35

hugging-face · seit 21. April 2026 · 12× · zuletzt 30. Juni 2026

Momentum

ML Intern (ml-intern) is an open-source agentic framework by Hugging Face, built on the smolagents framework, available as a CLI tool and as a web app/HF Space. The agent autonomously researches scientific papers (arXiv, HF Papers), searches and cleans datasets via the HF Hub, launches training runs (locally or via HF Jobs), and iterates on evaluation results for up to 300 steps. It was demonstrated on the PostTrainBench benchmark, improving a Qwen3-1.7B model from a 10% to 32% GPQA score in under 10 hours. The source code is released under the Apache-2.0 license on GitHub and requires a Hugging Face token (and optionally a GitHub token) for full functionality.

Historique du momentum

04.04.03.07.

Fonctionnalités

Autonomy Level	Fully autonomous agentic loop with up to 300 iterations (configurable via --max-iterations), including approval steps for critical actions
Channels	CLI terminal, web app/HF Space (also usable on mobile), Slack notifications (outbound only)
License	Apache License 2.0
Modalities	Text/code (reads papers, writes and trains ML code); primarily text-based agent control
Platform	CLI (local via uv) and web/mobile app as a Hugging Face Space
Price	Open-source/free; usage requires own API keys (Anthropic/OpenAI) or HF token, hosted inference billed to the active HF user
Release Date	April 21, 2026 (official launch)
Tool/MCP Integrations	HF Hub/Datasets/Jobs, HF Inference Providers, GitHub search, Trackio (experiment tracking), Slack notifications, local models via Ollama/vLLM/LM Studio/llama.cpp (LiteLLM)

Preuves (12)

Unternehmens-Analyse: Hugging Face

Stand 15.4.2026

SELLSynthszr Vote

On a risk‑adjusted basis, the balance of evidence favors a SELL stance at or near the last implied ~$4.5–5.0B valuation. Hugging Face’s strategic importance, strong developer network effects and rising Google Trends interest support a compelling long‑term narrative, but current valuation multiples (among the highest in the AI sector), modest and partially opaque revenue base, emerging security and governance risks, and recent layoffs tied to a business‑model pivot all point to an unfavorable near‑ to medium‑term risk‑reward profile. For new capital, better‑priced AI infrastructure or application names offer more attractive entry points; existing investors should consider partial liquidity where possible while monitoring for improved disclosure, clearer monetization traction and de‑risking of security and regulatory issues before adding exposure. (aventis-advisors.com)

Key Takeaways

Hugging Face remains a core infrastructure layer for open‑source and enterprise AI, with more than 2 million models on its hub, tens of millions of daily model downloads and views, and millions of registered users and contributors, underscoring strong network effects and high developer lock‑in despite intensifying competition from closed and open providers alike. (aiworld.eu)
The company’s last disclosed primary valuation is about $4.5–5.0 billion from its August 2023 Series D and subsequent market references, implying a very high EV/revenue multiple (~150x on 2023 ARR in one recent AI valuation study), which leaves little margin for error if growth or monetization underperform versus leading AI peers. (en.wikipedia.org)
Revenue is still modest relative to its strategic importance: various recent market data reports and investor write‑ups suggest ARR in the ~$50–130M range with expectations of ~$150M in 2024–2025, implying strong double‑digit to low‑triple‑digit growth but from a small base and with evolving mix toward APIs, managed inference and enterprise subscriptions. (sacra.com)
Hugging Face has begun to rationalize costs and pivot away from lower‑margin consulting toward recurring revenue streams, as evidenced by ~4% staff layoffs in early 2025 focused on enterprise services, while continuing to invest heavily in R&D and infrastructure to maintain its position as the default open‑source AI hub. (theinformation.com)
Security, governance and platform‑risk issues are emerging as key challenges: recent research and regulatory discussions highlight vulnerabilities in model serialization and exposed API tokens on the platform, as well as broader concerns about moderating model marketplaces—factors that could increase compliance costs and affect enterprise adoption if not managed proactively. (arxiv.org)

Action-Ideen

SELL

For investors with access to secondary shares at or above the last implied ~$4.5–5.0B valuation, risk‑reward now skews unfavorably: Hugging Face trades at one of the highest revenue multiples in the AI stack (~150x EV/Revenue vs. 20–60x for many other leading AI names), while monetization is still early, competition from hyperscalers and other open‑source hubs is intensifying, and recent layoffs plus security headlines introduce execution and reputational risk. De‑risking exposure here in favor of better‑priced AI infrastructure or application plays appears prudent.

Horizont: 12 Mon.

BUY

For long‑term, high‑risk investors who can access shares at a meaningful discount to the last round (e.g., via distressed secondary sales or structured deals), Hugging Face offers asymmetric upside as the de‑facto neutral hub for open‑source models: developer adoption, model count and traffic continue to grow exponentially, and the pivot toward APIs, managed inference and enterprise tooling could unlock operating leverage and higher‑quality recurring revenue over time, especially if the company successfully addresses security and governance concerns.

Horizont: 60 Mon.

HOLD

Existing investors who participated in or near the 2023–2024 rounds may be best served by holding: fundamentals (usage, community, enterprise traction) are still improving, and the platform remains strategically important to multiple hyperscalers and model providers, but elevated valuation, recent layoffs and security issues argue against aggressively adding exposure until there is clearer evidence of scalable, high‑margin monetization and improved risk controls.

Horizont: 24 Mon.

Google Trends · ↗ steigend

Global Google search interest for the term "Hugging Face" over the last two years shows a generally rising pattern with several pronounced spikes corresponding to major funding, partnership and security‑related news. Academic and policy analyses that track both website traffic and Google Trends indicate that interest in Hugging Face nearly doubled from late 2022 into 2024, and more recent data suggest continued elevated attention as the platform’s model hub surpasses two million models and becomes central to open‑source AI debates. The curve is volatile—reflecting news‑driven surges—but the baseline level of search activity today is materially higher than two years ago, consistent with its growing role in the AI ecosystem.

Contrarian Insights

• Despite widespread perception that Hugging Face is primarily a community‑driven, low‑monetization open‑source hub, recent data suggest a meaningful and growing enterprise and infrastructure business—projected ARR in the low‑hundreds of millions, 2,000+ dedicated enterprise clusters, and strong growth in inference services—indicating that, if priced reasonably, the company may be closer to a viable standalone SaaS/infrastructure business than many skeptics assume. (worldmetrics.org)
• While some investors view the 4% staff reduction and shift away from bespoke consulting as a sign of weakness, it can also be interpreted as early discipline: reallocating resources from low‑margin services to scalable APIs and subscriptions may improve unit economics and make Hugging Face more resilient than peers that continue to rely heavily on custom enterprise work to justify high valuations. (theinformation.com)