Key Takeaways

  1. Amsterdam-based AI cloud company Nebius (NASDAQ: NBIS, market cap ~$23.3B) has agreed to acquire Tel Aviv-founded startup Tavily for an initial $275 million in cash, with the total potentially reaching $400 million pending performance milestones.
  2. Tavily built the purpose-built real-time search engine for AI agents, powering over 1 million developers and serving Fortune 500 clients including IBM, Cohere, and Groq.
  3. Founded in late 2024 with only $25 million in total funding, Tavily’s exit ranks among the fastest in Israel’s AI scene, turning a ~$121M post-money valuation into a deal worth up to $400M in under 18 months.
  4. The agentic AI market will grow from $7.29 billion in 2025 to $139.19 billion by 2034 at a 40.5% CAGR, making search infrastructure for autonomous agents a critical capability for enterprises.

Quick Recap

On February 10, 2026, Nebius Group, the Amsterdam-headquartered AI cloud company carved out of Russian tech giant Yandex, officially announced a definitive agreement to acquire Tavily. Which is an Israeli agentic search startup. The deal, first reported by Calcalist, involves an upfront payment of $275 million in cash, with the potential for an additional $125 million in cash or Nebius shares contingent on performance milestones. Tavily’s 30-person team, including founder and CEO Rotem Weiss, will join Nebius and continue operating under the Tavily brand.

Inside the Deal: Infrastructure Play, Not Just an Acqui-Hire

This acquisition is far more than a talent grab. Nebius is pursuing a strategic vision to become a vertically integrated, full-stack AI cloud platform — and Tavily fills a critical gap in that stack.​

Nebius already operates its own data centers and GPU infrastructure across Europe and the U.S., with several hundred megawatts of total capacity. In September 2025, Microsoft became a major Nebius client under a multiyear $17 billion revenue agreement for computing capacity. The company’s inference product, Token Factory, lets enterprises run open-source AI models on Nebius infrastructure. By integrating Tavily, Nebius gains a production-grade search layer that allows AI agents built on its platform to access real-time, structured web data without relying on third-party vendors.

Tavily’s technology is specifically engineered for autonomous AI systems. Unlike traditional search engines that return noisy HTML pages with ads and irrelevant content, Tavily’s API delivers cleaned, ranked, and LLM-optimized snippets in a single call. The platform supports five core endpoints — search, extract, crawl, map, and research — and has achieved over 3 million monthly SDK downloads. Its clients include IBM, Cohere, and Groq, and it integrates natively with LangChain, LlamaIndex, Azure MCP, and Snowflake.

Roman Chernin, Nebius co-founder and CBO, framed the acquisition clearly:
“We’re not just an infrastructure-as-a-service company — we’re building the complete platform for anyone who wants to build AI products, agents, or services.”

Why Now: The Agentic AI Inflection Point

The timing of this deal is no coincidence. The AI industry is undergoing a structural shift from chatbots and assistants toward fully autonomous agents that plan, reason, and take action independently. Industry estimates cited by Nebius suggest that AI agents are expected to generate more internet queries than humans within the next few years.​

The agentic AI market is projected to surge from $7.29 billion in 2025 to $139.19 billion by 2034, growing at a 40.5% CAGR. Within this market, search infrastructure has emerged as a critical enterprise capability — agents that cannot access real-time data are fundamentally limited in their usefulness.

Tavily’s founder Rotem Weiss put it bluntly:

“Tavily is on a mission to onboard the next billion AI agents to the web.”

For Israel’s tech ecosystem, this deal underscores a broader trend. The country is increasingly exporting not just cybersecurity but core AI infrastructure. Nebius was recently selected by the Israel Innovation Authority to build the country’s national AI supercomputer, equivalent to 1,000 NVIDIA B200 accelerators. Tavily’s rapid rise — from founding to a $400M exit in roughly 18 months — signals that Israeli startups are moving from the periphery to the core of the global AI stack.

Competitive Landscape & Comparison

Tavily operates in the fast-growing agentic search API market alongside several notable competitors. The two most comparable rivals are Exa (San Francisco, $700M valuation, $107M raised) and Linkup (Paris/NYC, $10M seed round, early stage).

Feature / MetricTavily (Nebius)ExaLinkup
FoundedLate 2024 (Tel Aviv)​2021 (San Francisco)​2024 (Paris)​
Total Funding$25M (pre-acquisition)​$107M (Series B)​$10M (Seed)​
Valuation / DealUp to $400M (acquisition)​$700M (last round)​Undisclosed
Developer Reach1M+ developers, 3M+ monthly SDK downloads​Not publicly disclosedThousands of clients​
Core ApproachCustom search algorithm, LLM-optimized snippets​Semantic / neural search via own index​Own web index with sub-second API​
Latency~210ms typical, 90ms ultra-fast mode​Sub-200ms (Exa Instant)​Sub-second (/fast endpoint)​
People / Company SearchNo​Yes (dedicated endpoints)​No
Enterprise ClientsIBM, Cohere, Groq​Not publicly listedArtisan, KPMG​
Agentic CapabilitiesSearch, extract, crawl, map, research endpoints​Search API + Websets for deep queries​Search + retrieval APIs​
Key DifferentiatorNow part of Nebius full-stack AI cloud​Independent; semantic retrieval + benchmarks lead on complex queries​Fastest sub-second accuracy; early stage​

Exa leads in benchmark performance on complex multi-hop retrieval tasks (81% vs. Tavily’s 71%) and offers unique people and company search capabilities. However, Tavily’s integration into Nebius’s full-stack AI cloud — with GPU infrastructure, Token Factory inference, and now search in one platform — creates a bundled value proposition that standalone API providers cannot easily match. Linkup is the earliest-stage competitor but its sub-second accuracy focus and backing from Mistral-adjacent investors position it as one to watch.

Sci-Tech Today’s Takeaway

I think this deal is a genuinely big deal — and not just because of the eye-popping $400M price tag for a company barely 18 months old. In my experience covering AI infrastructure, the acquisitions that matter most are the ones that collapse vendor complexity. That’s exactly what Nebius is doing here. By bringing Tavily’s search layer in-house alongside its GPU cloud and inference engine, Nebius is building the kind of one-stop AI platform that enterprise buyers increasingly demand. I generally prefer to be cautious about “full-stack” narratives — they often over-promise — but Nebius has the receipts: a $17B Microsoft deal, a national AI supercomputer contract, and now the most widely adopted agentic search API in its toolkit.

For developers, this is bullish. The fragmented tool-chain of stitching together search, inference, and compute from three different vendors has been a real friction point. If Nebius executes the integration well, it could meaningfully lower the barrier to deploying production-grade AI agents. For the broader market, this deal validates agentic search as a standalone, multi-billion-dollar category — not just a feature tacked onto existing search engines. Exa’s $700M valuation and Linkup’s fresh seed round confirm the trend. My verdict: watch the “search for AI” space closely. It’s where the next wave of infrastructure winners will emerge

Add Sci-Tech Today as a Preferred Source on Google for instant updates!
google-preferred-source-badge
Joseph D'Souza
(Founder)
Joseph D'Souza founded Sci-Tech Today as a personal passion project to share statistics, expert analysis, product reviews, and experiences with tech gadgets. Over time, it evolved into a full-scale tech blog specializing in core science and technology. Founded in 2004 by Joseph D’Souza, Sci-Tech Today has become a leading voice in the realms of science and technology. This platform is dedicated to delivering in-depth, well-researched statistics, facts, charts, and graphs that industry experts rigorously verify. The aim is to illuminate the complexities of technological innovations and scientific discoveries through clear and comprehensive information.