Top 10 AI startups that raised over $100M in 2025

AI fatigue is real for most of us, but surprisingly still not for investors. I have written about the top 10 AI startups that raised rounds of $100 million or more in 2024. And in reality, there are 49 AI startups in the US alone, which sounded outrageous. But 2025? Well, it’s on track to go even bigger, and there are noticeably more companies raising multiple mega-rounds within the same year.

So, let’s unpack the top 10 AI startups that raised over $100M in 2025, ranked from the biggest rounds down, and see what they’re actually building behind those headline numbers.

№10 – Sierra — $350 Million

Customer service is one of the first real enterprise functions being seriously reimagined by AI.

Sierra, founded by Bret Taylor (ex-CEO of Salesforce and co-creator of Google Maps), is one of the startups leading that charge.

In September, Sierra raised $350 million in a round led by Greenoaks Capital, valuing the company at more than $10 billion. Its platform builds AI agents that can handle complex customer workflows while still respecting enterprise-grade guardrails.

The bet here is clear: as AI agents become more capable, companies won’t just use them to answer FAQs—they’ll let them handle refunds, escalations, and eventually entire processes. Sierra wants to be the control layer that enterprises actually trust.

№9 – SandboxAQ — $450 Million

AI isn’t just about language and images. It’s also colliding with physics, cryptography, and quantum-inspired math.

SandboxAQ, which builds AI-powered systems for fields like cybersecurity, simulation, and computational chemistry, closed a $450 million Series E in April at a $5.7 billion valuation. The round included Nvidia, Google, and Ray Dalio, among others.

The company sits at the intersection of AI plus quantum-inspired algorithms, plus high-performance computing, targeting heavily regulated, high-stakes sectors like defense, pharmaceuticals, and finance.

If “traditional” AI startups are racing for user scale, SandboxAQ is quietly positioning itself where contracts are slower—but the tickets are much, much larger.

№8 – Lambda — $480 Million

If you’ve ever tried to spin up an LLM on a GPU and hit a waitlist, you’ve felt the problem Lambda is solving.

In February, the AI infrastructure company raised a $480 million Series D at a roughly $2.5 billion valuation. The round was co-led by SGW and Andra Capital, with Nvidia, ARK Invest, and others participating.

Lambda sells GPU clusters, cloud infrastructure, and on-prem boxes tuned specifically for AI workloads. They’re the middle layer between raw silicon and the teams training models—and they’ve quietly become a go-to provider for both startups and enterprises who can’t get enough capacity from hyperscalers alone.

When everyone needs GPUs, the companies that know how to run, rent, and optimize them become very valuable.

№7 – Groq — $750 Million

If 2023–2024 belonged to the GPU, 2025 is making a case for alternative inference hardware.

Groq, an AI inference company that went viral with its “lightening fast” LLM demos, raised a $750 million in Series D-3 in September at a $6.9 billion valuation. The round was led by Disruptive.

Groq’s value prop is simple: run AI models at high speed and low latency, especially for workloads that can’t afford delay, such as financial trading, real-time agents, and interactive apps. Where Nvidia dominates training, Groq wants to own the inference lane.

Investors seem to believe there’s room for both.

№6 – Cerebras Systems — $1.1 Billion

Not every AI winner is a model company. Some are building the picks, shovels, and silicon.

Cerebras Systems, known for its wafer-scale AI chips, raised a $1.1 billion in Series G at an $8.1 billion valuation in September. The round was co-led by Fidelity and Atreides Management.

Cerebras’ chips are designed to train massive models faster by putting an entire compute cluster onto a single piece of silicon. As model sizes grow, architectures like this look less like experiments and more like necessities.

In a world where GPUs are the new oil, Cerebras is one of the few companies trying to build an entirely different engine.

№5 – Thinking Machines Lab — $2 Billion

Some funding rounds make you blink twice. Thinking Machines Lab is one of them.

The AI research lab, founded by former OpenAI CTO Mira Murati, raised a $2 billion seed round in July at a $12 billion valuation. Yes, you read that correctly: $2B. Seed. $12B valuation.

Their mission? To push toward human-level reasoning rather than just pattern completion. Backers include Andreessen Horowitz, Nvidia, Accel, and AMD, signalling a rare alignment across top-tier VCs and the GPU giants themselves.

If OpenAI was “ACT I” of frontier labs, Thinking Machines is positioning itself as “ACT II.”

№4 – Reflection AI — $2 Billion

China’s DeepSeek proved that open, aggressively optimized models can terrify both Big Tech and regulators.

Enter Reflection AI, a U.S.-based competitor building low-cost, open, fully American alternatives. In October, the company raised a massive $2 billion Series B led by Nvidia, at an $8 billion valuation—its second huge round of the year.

Reflection AI isn’t just building models; it’s building a geopolitical hedge. The goal: make sure U.S. and allied countries have domestic, open models that can keep pace with DeepSeek-style innovations without relying on external ecosystems.

This isn’t just a tech story. It’s industrial policy wrapped in an AI startup.

№3 – Anysphere — $2.3 Billion

2025 is the year AI coding tools stopped being gimmicks and started getting real money.

Anysphere, the AI research lab behind Cursor, the “vibe coding” editor that has gone viral with developers, raised a $2.3 billion round in November at a $29.3 billion valuation. This was their second mega-round of the year, on top of a $900 million Series C in June that valued them near $10 billion.

Cursor’s pitch is simple but powerful: build a coding environment where AI feels native, not bolted on. Instead of copy-pasting from ChatGPT, developers stay inside their editor while the AI writes, refactors, and reasons about entire codebases.

If GitHub Copilot was the warm-up, Anysphere is trying to be the operating system for AI-era software engineering.

№2 – Anthropic — $13 Billion

If OpenAI is the AI giant, Anthropic is the disciplined rival quietly stacking war chests.

In September 2025, Anthropic announced a $13 billion Series F at a $183 billion valuation—its second multi-billion-dollar raise of the year, after a $3.5 billion Series E in March that valued the company at $61.5 billion.

Best known for its Claude models, Anthropic’s pitch is a mix of “safer AI” and enterprise-ready foundation models. With a total of $16.5 billion raised in just one year, Anthropic isn’t just keeping up. It’s building the kind of capital base normally reserved for national infrastructure projects.

If you think this is just about chatbots, you’re not paying attention.

№1 – OpenAI — $40 Billion

Let’s start with the elephant in the GPU room.

In March 2025, OpenAI raised a record-breaking $40 billion round at a jaw-dropping $300 billion valuation. The round was led by SoftBank, with Microsoft, Thrive, and Coatue piling in.

This isn’t just another funding round! It’s a geopolitical event disguised as a term sheet. OpenAI is racing to stay ahead in large language models, multimodal AI, and whatever comes after GPT. The capital gives them firepower for compute, research, and their increasingly ecosystem-like strategy: models, tools, infrastructure, and partnerships.

If you wanted a single data point that says “AI isn’t slowing down anytime soon,” this is it.

So… Are These AI Startups Overhyped?

When you see numbers like $40 billion, $13 billion, and $2 billion seed rounds, it’s natural to ask: have investors lost their mind?

The truth is, most companies on this list will justify their valuation. Some may flame out, sometimes facing such serious competition from tech giants, as recent OpenAI declared ‘code red’, facing Google Gemini 3 is an example. Others will likely pivot three times before finding product–market fit. 

That’s the nature of venture capital: a handful of outliers pay for a long tail of failed bets.

But zoom out, and the pattern is clear:

AI Infrastructure is being aggressively funded for AI model research by companies such as OpenAI and Anthropic.  Meanwhile, alternative hardware (Groq, Cerebras, Lambda) is no longer niche. 

We’re watching investors place enormous, concentrated bets on the belief that AI will be as foundational as the internet in the 90s and that whoever owns the models, infrastructure, and workflows will shape the next few decades of value creation.

I’m usually reluctant to make predictions about technology, but I feel fairly confident about this one: in a couple decades there won’t be many people who can write.  — Paul Graham, from YC on October 2024

Looking forward 

Whether that plays out as expected is still an open question.

But one thing is certain: If 2025 is any indication, the AI funding story is far from over. And this top 10 list? It’s probably just the prologue.

If you enjoy learning and entrepreneurship, you can follow me on my YouTube channel or my newsletter, I’m here to practice my entrepreneurship muscle every week! Stay tuned, and see you in the next one!