While tech giants like OpenAI and Meta dominate the headlines, a new breed of companies is rising rapidly, building foundation models, AI infrastructure, and end-user applications that are reshaping traditional industries. Some directly challenge the dominance of Big Tech, while others quietly embed themselves deep within the AI value chain.
In 2025, Generative AI startups collectively raised over $69.6 billion, accounting for 11.77% of all venture capital deployed worldwide. Yet as capital pours in, the picture is far from uniform. The United States dominates foundation model development, but Europe is fast emerging as a serious contender — not by mimicking Silicon Valley’s playbook, but by leaning into its strengths: technical depth, regulatory foresight, and vertical specialization.
Vestbee unpacked the current state of GenAI venture capital landscape, mapping global investment trends, zooming in on Europe, and spotlighting the biggest funding rounds and most notable startups across three key segments of the AI stack: model makers, infrastructure, and applications.
GenAI in VC: globally
The AI investment boom is not cooling off. In 2024, Generative AI startups raised over $48 billion, a twentyfold increase from 2020. That’s nearly as much as all venture capital raised in Europe during the same year. And the momentum has only intensified: by mid-2025, AI companies globally had already secured $69.6 billion in funding, representing 11.77% of all VC deployed across sectors.
What began as a bet on foundation models has now matured into a full-stack investment opportunity. While over 50% of GenAI funding has gone to model makers, such as OpenAI, Anthropic, and Mistral, investor attention is increasingly turning to applications (19%), apps with proprietary models (17%), and the operation layer (9%), where defensibility and distribution are often easier to establish.
Applications. The second-largest funding category spans a wide range of verticals and media types — from chat-based assistants and internal copilots to tools for code, design, and 3D asset creation. Notably, text-based use cases still dominate, including copywriting, customer support, and enterprise knowledge systems. Applications typically fall into two camps: those building on proprietary models, and those leveraging third-party APIs.
Operations layer (Infra). As GenAI adoption scales, so does the need for purpose-built tooling. From prompt engineering frameworks and vector databases to MLops platforms, infrastructure has become a critical enabler. This includes players like Pinecone and Weaviate, who helped push vector DB funding to an all-time high of $210 million in 2023, and legacy MLops providers like Scale AI, now layering in GenAI functionality. Major cloud providers and Big Tech firms are also racing to build custom AI hardware to reduce dependency on NVIDIA and improve cost margins.
When it comes to round sizes, in 2024 GenAI companies consistently outpaced both broader AI and general tech peers — often by huge margins, showing a significant premium at every stage, from Seed to Series C+.
- At the seed stage, GenAI startups raised an average of $8.9 million, nearly double the average for traditional tech ventures and that of other AI startups ($4.8 million). The gap widened further at Series A, with GenAI rounds reaching $34 million — a full 70% above broader AI ($20 million).
- By Series B, GenAI’s dominance became even more pronounced: $199 million on average, nearly three times the broader AI benchmark and four times the rest of tech. Even at Series C and beyond, where deal sizes often plateau, GenAI companies raised an average of $213 million — nearly twice the amount other tech startups secured.
GenAI in VC: Europe
From 2019 to 2024, European GenAI startups raised a total of $6.1 billion, a fraction of the global total but still a significant signal of the continent’s growing momentum in the field.
In a landscape dominated by US and Chinese giants, Europe is positioning itself as a strategic AI contributor, particularly in open-source models, enterprise-grade applications, and ethical AI tooling. France has emerged as the regional frontrunner, producing all three of Europe’s notable frontier models in 2024 (Mistral, Kyutai, and LightOn). Compared to the US, European AI startups get smaller rounds, often aimed at technical depth, research spinouts, or B2B infrastructure plays.
There is also a more pronounced public-private synergy, with initiatives like France’s €500 million AI fund (via Bpifrance) or Germany’s AI startup ecosystem supported by federal R&D programs. Many of Europe’s GenAI startups are also leveraging open science ecosystems and university partnerships - for instance, Hugging Face (originally French) and Aleph Alpha (Germany) are both built on strong academic-research roots.
Beyond France, when it comes to geography, other emerging hubs of AI innovation include Germany (especially in mobility, robotics, and manufacturing), the UK (prominent synthetic data platforms and generative design tools) and the CEE region (a fast-rising source of technical talent and lean innovation, with increasing VC interest from pan-European and US funds).
Despite its growing momentum, Europe’s AI ecosystem still faces several structural troubles. According to Dealroom, scaling remains a persistent challenge, as many promising startups struggle to secure late-stage funding domestically and often turn to US or Asian investors to bridge the gap. Access to advanced compute infrastructure is another limiting factor — most European startups continue to rely on US-based cloud providers, which raises both cost and sovereignty concerns.
At the same time, the continent continues to experience a steady talent drain, with top researchers and engineers often lured by the scale, compensation, and opportunity of labs and startups in the US. These constraints don’t neccesarily undermine Europe’s position, but they do slow the continent’s ability to compete at the scale and speed set by global leaders.
Most notable global funding rounds in AI
- OpenAI remains the undisputed leader, pulling in a massive $40 billion round in early 2025, led by Microsoft, SoftBank, Founders Fund and Magnetar Capital, pushing its valuation to $300 billion.
- Elon Musk’s xAI raised $10 billion at an $80 billion valuation pre-raise, with backers including Morgan Stanley and other undisclosed institutional players. The company also closed two additional $6 billion rounds within a year — bringing its total to $22 billion — with support from investors such as Andreessen Horowitz, Sequoia Capital, and Qatar Investment Authority.
- Another standout is Scale AI, which raised a staggering $14.3 billion at a $29 billion valuation, backed by Meta, doubling down on its position as a leading data infrastructure provider for AI models.
- Similarly, Anthropic, the team behind Claude, continued its funding spree with a $3.5 billion injection that brought its valuation to $61.5 billion — thanks to Salesforce Ventures, Lightspeed and Alphabet— alongside a separate $4 billion commitment from Amazon as part of its broader strategic partnership.
- Notable infrastructure-focused players also drew attention: CoreWeave, an AI cloud compute startup, secured $1.1 billion at a $19 billion valuation, and Groq, known for its ultra-low latency AI chips, brought in $1.5 billion from the Kingdom of Saudi Arabia.
- Meanwhile, Lambda, a training-focused GPU cloud company, raised $480 million with participation from NVIDIA and ARK Invest.
The generative AI space wasn’t limited to American giants.
- Mistral AI and Aleph Alpha in Europe, as well as Moonshot AI in China, joined the unicorn club with billion-dollar rounds. Even in defence and embodied AI, the capital kept flowing.
- UK-based Wayve secured $1.1 billion from SoftBank, Microsoft and NVIDIA to further its self-driving tech, while US defense tech company Anduril Industries raised $1.5 billion to scale autonomous military systems.
- Perhaps most intriguing is the rise of AI-first biotech like Xaira Therapeutics, and experimental labs like Safe Superintelligence, both of which raised rounds in the $1–2 billion range, signaling VC appetite even for highly speculative moonshots.
In Europe, according to State of Tech 2024, other major rounds that year outside of Mistral AI reflected the growing diversity of the European AI scene:
- Helsing raised €450 million for defence AI;
- Poolside AI closed a €400 million round to scale its software engineering foundation models;
- DeepL brought in €300 million to advance its neural translation technology.
- Emerging players included Magic.dev ($117M), originally founded in Austria with ambitions for general-purpose AI in code generation, and CEE-founded and UK-based ElevenLabs, which raised $80 million for multilingual voice AI.
Case studies
We examined nine standout startups — from Paris-based Mistral AI, racing to make Europe a global AI contender, to Zhipu AI, China’s state-backed answer to GPT-4, to Synthesia, whose synthetic avatars are now speaking in 140 languages for Fortune 100 companies and ElevenLabs, the Warsaw- and London-based startup whose AI voices are quietly becoming used for everything from audiobooks to enterprise agents.
AI models
Open AI
OpenAI stands at the center of the AI ecosystem, with its immense funding and long-standing partnership with Microsoft reinforcing its leadership in the field. Strategically, OpenAI is pursuing a full-stack approach: developing frontier models, expanding deployment infrastructure through partnerships, and broadening its commercial offerings (notably enterprise Copilot products).
In mid-2025, it released GPT-4.1 and GPT-4.1 mini for ChatGPT, which deliver faster inference and better coding performance than previous versions. The company is also expanding multimodal capabilities (voice, image, video) via GPT-4o and related research. OpenAI’s partnerships and initiatives remain high-profile: it is planning new data-center builds (e.g. with SoftBank) and has reportedly acquired AI coding startup Windsurf for ~$3 billion (Bloomberg) to boost its tooling.
Funding
OpenAI raised massive capital to scale its models. In 2025 it announced a $40 billion funding round (led by SoftBank) that values the company at roughly $300 billion. SoftBank contributed about 75% of the funding, with the balance from investors including Microsoft, Thrive Capital, Altimeter, and Coatue Management. In total, OpenAI has raised ~$63.9 billion since its inception
Mistral AI
Mistral AI is a Paris-based startup building open-weight large language models. Its goal is to create cutting-edge generative AI (text, code, etc.) and provide Europe and global customers an alternative to U.S./Chinese LLMs. From launch, Mistral’s models (e.g. a 7B-parameter “Mistral” model) have garnered attention for strong performance. Co-founder Arthur Mensch emphasizes open-source principles (“open weights”) and high inference speed. The French government has actively promoted Mistral as a strategic homegrown AI asset. Strategically, Mistral helps diversify the AI landscape and may partner with cloud or device vendors to compete on cost/performance outside North America.
Mistral has been actively expanding its product lineup and partnerships. In February 2025, it launched Le Chat, a multilingual AI chatbot app on web and mobile, available as native iOS and Android launches. The app boasts ultra-fast inference—up to 1,000 words per second—positioning it as a Europe‑based alternative to ChatGPT.
Funding:
In June 2024, it closed a €600 million Series B round led by General Catalyst, valuing the company at approximately €5.8 billion and pushing its total funding above €1 billion. Investors included Andreessen Horowitz, Lightspeed, Cisco, Microsoft (which had previously invested $16 million), and Nvidia
Zhipu AI
Zhipu AI, spun out of Tsinghua University in 2019, is widely viewed as China’s closest equivalent to OpenAI. Backed by both state-affiliated and private capital, it has emerged as one of the country’s leading “AI Tigers” — a term now used to describe China's most competitive foundation model players, alongside Moonshot AI, Baichuan, DeepSeek, and others. Zhipu focuses on Chinese-language optimization, agentic model development, and sovereign AI infrastructure integration. Its flagship model family, GLM (General Language Model), powers a growing suite of open-source tools for both consumers and enterprise deployment. In 2025, Zhipu launched GLM‑4.5, a next-generation model designed for agentic reasoning and coding use cases, while continuing to support GLM‑4 Voice, its speech-enabled large language model. The company also announced plans to open-source a comprehensive suite of models throughout the year, covering foundation models, multimodal systems, and agent-based frameworks. As part of its regional expansion strategy, Zhipu is partnering with the City of Chengdu to develop “Zhipu Zhuge,” a localized LLM tailored to the city’s linguistic and policy context, alongside the creation of a public-private AI training center. The company operates at scale across national supercomputing clusters and is deeply embedded in China’s state-backed AI infrastructure zones, including Zhongguancun and Chengdu.
Zhipu AI’s trajectory reflects China’s sovereign AI ambitions in sharp relief: state-aligned, open-source-friendly, and focused on national infrastructure integration.If Zhipu can maintain its pace of technical innovation—particularly in multimodal and agentic systems—it may emerge not just as a national champion, but as a viable global counterweight to Western model giants.
Funding:
In december 2024, Zhipu raised ¥3 billion (~$412M) from state-linked investors, including Zhongguancun Science City and funds in Hangzhou, and in early 2025, it secured an additional ¥300 million (~$42M) from the Chengdu municipal government to co-develop a regional AI initiative and training hub. The valuation is estimated at ¥20 billion (~$2.7B) in mid-2024, with more recent estimates suggesting it may have climbed to ¥40 billion (~$5.6B) after the latest funding rounds. (Exact valuation figures remain unofficial).
Applications
Character.AI
The U.S.-based startup offers a conversational AI platform where users create and interact with custom “characters”—chatbots with distinct personas. Unlike single-assistant bots, Character.AI is built around role-play and storytelling, enabling users to chat with AI versions of historical figures, fictional characters, or community-generated creations. Character.AI experienced explosive growth after opening to the public in September 2022, quickly reaching 100 million monthly site visits, with average engagement exceeding 2 hours per day per messaging user. The platform’s design niche - blending entertainment, creativity, and a unique personal touch differentiates it from more utilitarian-like chatbot apps
Funding:
In March 2023, Character.AI closed a $150 million Series A led by Andreessen Horowitz (a16z), with participation from Nat Friedman, SV Angel, Elad Gil, and A Capital. The round valued the company at approximately $1 billion, despite having no revenue at the time
Eleven Labs
Eleven Labs was founded in Poland and though it is now registered in New York, it still retains its strong European roots. The startup specialises in highly realistic speech synthesis, voice cloning, and AI dubbing, aiming to make synthetic speech virtually indistinguishable from real human voices across languages, accents, and emotional tones. It has established itself as a global leader in voice AI, with rapid product offering expansion across recent years - from mobile voice apps to publishing platforms and enterprise-grade conversational agents. In June 2024, ElevenLabs launched "Reader", a platform where users can upload any text or document and listen to it read in a human-quality voice — a clear play for the audiobook and education markets. The firm has also added podcast-generation (GenFM) and audiobook publishing features with integrated monetization for rights holders. ElevenLabs has also formed partnerships with publishers (e.g. The Washington Post, The Atlantic, MIT Technology Review) and gaming studios such as Paradox Interactive and Cloud Imperium Games. Strategically, ElevenLabs is well-positioned to own the synthetic voice and localization niche within the broader content ecosystem.
Funding:
In June 2023, ElevenLabs raised $19 million in a Series A co-led by Andreessen Horowitz, Nat Friedman, and Daniel Gross, at a valuation of about $100 million. Later, in January 2024, it completed an $80 million Series B round co-led by the same investors along with Sequoia Capital, raising its valuation to approximately $1.1 billion, making it a unicorn. As of January 2025, ElevenLabs closed a $180 million Series C led by a16z and ICONIQ Growth, with participation from NEA, World Innovation Lab, Valor, Endeavor Catalyst Fund, and Lunate, bringing the total funding to approximately $281 million and valuing the company at around $3.3 billion.
Synthesia
Synthesia is one of the most prominent enterprise-facing generative AI video companies, showcasing how realistic synthetic videos can be adopted on a large scale by enterprises. The company has pioneered an AI-driven video creation platform that transforms text into realistic talking-head videos, powered by avatars and voice synthesis - allowing firms to create internal training, educational content, or marketing videos at scale - without the need for live filming or actors. By early 2025, Synthesia reported serving over 60,000 customers, including more than 60% of Fortune 100 companies, with clients spanning Reuters, Zoom, Microsoft, and Xerox. In the product lane, Synthesia introduced interactive video templates, personal avatar creation (via selfie uploads), real‑time teleprompter functionality, and deeper integration with learning management systems (LMS) for corporate training.
Funding:
As of January 2025, Synthesia has raised a total of $330 million in equity. Its most recent was a $180 million Series D led by NEA, with participation from GV (Google Ventures), MMC Ventures, World Innovation Lab, Atlassian Ventures, and PSP Growth. This round valued the company at approximately $2.1 billion, making it the UK’s most valuable AI media startup.
