February isn't over, and the numbers are already staggering.

ElevenLabs $500M. SkildAI $1.4B. humans& $480M. Anthropic $30B. In roughly seven weeks — from January through mid-February — US-based AI startups have closed more than 17 rounds of $100M or more. Compare that to 55 for all of 2025, and the pace shift is unmistakable.

But the headline numbers only tell half the story. To understand what's actually happening, you have to look at where the money is going — and what these companies are genuinely building.

The Deal Landscape — Key Rounds at a Glance

Here's how the top 2026 funding rounds stack up so far.

Notice the mix: some of these companies have real ARR (Annual Recurring Revenue), others have nothing yet. That distinction matters — a lot — and it's worth keeping in mind as you read the rest.

Three Capital Flows — Voice AI, Robotics, Research Labs

Zoom out and you can see three distinct channels drawing capital right now.

Flow 1: Voice AI

ElevenLabs turns text into human-quality speech and clones voices across languages — Deutsche Telekom is already deploying it for customer service at scale. ARR hit $330M, up more than 3x year-over-year, and the Series D was led by Sequoia, with a16z quadrupling its position. VCs doubling or quadrupling existing stakes is rare; it's about as strong a conviction signal as you'll see. Valuation went from $3.3B a year ago to $11B today.

Deepgram ($130M, $1.3B valuation) sits in the same lane — speech-to-text infrastructure for call centers and meeting transcription. Two major voice AI rounds closing in the same cycle isn't coincidence. It's a market-is-paying signal.

Flow 2: Robotics

SkildAI isn't building robots. It's building the AI brain that makes robots work — think iOS, not the iPhone. The company trains AI models that teach robots how to grasp objects, navigate, and execute physical tasks. SoftBank led the $1.4B round at a $14B valuation, with Nvidia and Jeff Bezos participating. That combination of investors — one with deep hardware roots, one extending its chip ecosystem — in a single deal signals that the robotics AI market has arrived in earnest.

Flow 3: Research Labs

Anthropic raised another $30B, valuing the company at $380B. Singapore's GIC led the round, with D.E. Shaw and Founders Fund among the participants. It's the second-largest single VC transaction in history. Revenue run rate is $14B, growing 10x annually for three consecutive years.

What These Companies Actually Do

A lot of the names making funding headlines are genuinely unclear on what they build. Here's the short version for the ones that keep coming up.

The Strangest Deal: humans& $480M (No Product)

This one deserves its own section.

humans& is three months old. No product. No revenue. It just closed a $480M seed round at a $4.48B valuation — the second-largest seed round in VC history.

The team explains part of it: an Anthropic RL researcher, two engineers who worked on Grok at xAI, Google's seventh employee, and a Stanford professor. Nvidia and Jeff Bezos are in the cap table. The stated mission is building AI that augments human capability rather than replaces it — though no specific product direction has been disclosed.

Reaction in the developer community was split. On Hacker News, the deal was described as "a case study in how far team-based betting can go." One camp argued that a team this credentialed could be trusted to build something worth backing at any valuation. The other maintained that $4.5B for a company with nothing shipped is simply indefensible. Both camps are still there.

Bubble or Breakthrough?

The AI sector burns around $400B a year while recovering only $50–60B. How should you read this investment wave against that backdrop?

The honest answer is: it's both, and they're not evenly distributed.

ElevenLabs has $330M in real ARR. Anthropic has been growing 10x annually for three straight years. That's not a bubble — that's actual enterprise adoption translating into durable revenue. Corporate customers are writing checks and getting measurable returns: ElevenLabs' B2B customers are seeing real cost reductions in call center operations and multilingual support automation.

At the same time, deals like humans& — $4.5B on team pedigree alone — coexist in the same market. Developer communities continue pointing out that most AI startups are thin wrappers on existing APIs with no real moat and no meaningful differentiation.

It's not a bubble across the board, and it's not entirely real across the board either. One VC on Hacker News put it well: "selective overheating" — worst at the pre-product seed stage, a different story entirely once there's a customer base and revenue.

What This Cycle Signals

If you had to summarize 2026 AI investment in one sentence: companies with real revenue are scaling faster, while companies with none are running on team reputation alone.

Voice AI, robotics, and AI infrastructure are attracting capital because they're already generating returns. Anthropic is a separate category — the scale and time horizon are different. The research-lab bets belong in their own column.

Two things will determine whether this wave was justified. First: what does humans& actually ship? Second: when does SkildAI's robotics brain reach real factory floors at scale? Those two outcomes will serve as the clearest stress tests for the current valuation environment.

It's easy to get lost in the raw numbers when AI headlines drop daily. But the most useful filter remains the same: is there real ARR behind the raise or not? A $480M raise with zero ARR and a $500M raise with $330M ARR are completely different bets — they just happen to look the same in the headline.