Greylock closes $1.5B Fund XVIII, backs ex-Palo Alto Networks founder's new cybersecurity startup SciLake with $36M opening check

Greylock closes $1.5B Fund XVIII, leads $36M check into SciLake

Greylock has closed its eighteenth venture fund at $1.5 billion, focused on early-stage AI companies. The fund's headline bet so far is SciLake, a new cybersecurity platform founded by Nir Zuk, who also founded Palo Alto Networks. Greylock's opening check into SciLake is $36 million.

The contrast with Palo Alto's original financing is striking. When Greylock backed Zuk's first company roughly 21 years ago, the initial check was $250,000. Palo Alto Networks now trades at nearly $300 billion in market cap. Motamedi's logic is explicit: when the founder has already built the largest company in cybersecurity history, the capital requirement at inception looks very different.

“We've got some big news today. We're announcing Greylock eighteen, our eighteenth venture fund. It's a billion and a half... The founder of that company, Nir Zhuk, left Palo Alto last year. He started a new company with Greylock. That company is called SciLake. It's building a really new important cybersecurity platform. Our initial check there is $36,000,000.”

How Motamedi maps the AI opportunity

Motamedi argues the firm is investing up and down the stack because the mistake in AI is underestimating the size of the wave, not picking the wrong layer. Greylock holds positions in Anthropic and OpenAI at the foundation model level, Baseten (inference infrastructure, backed in 2019 before generative AI was mainstream) at the infrastructure level, and a range of application companies including Resolve AI, which fully automates on-call incident response for customers like Coinbase.

On the application side, Motamedi points to December 2025 model releases as the inflection point that made robust long-running agents viable. Resolve and Seven AI (security operations) are his cited examples of products that are now commercially real because the underlying models can sustain the workflows.

Why bigger checks make sense even as costs fall

The fund is larger not because Greylock is abandoning early-stage discipline, Motamedi says, but because the ceiling on outcomes has expanded. He pushes back on the "capital-light startup" narrative, arguing that capital efficiency and capital deployment are not in tension: if you can do more with each dollar, you should deploy more dollars faster to widen the lead.

He points to Cresta, a customer support AI company Greylock seeded, as evidence of the multiple-winner dynamic. Cresta is now north of $100 million ARR and growing, with at least two other companies at similar scale in the same category. His claim is that AT&T alone spends $1 billion a year on contact center operations, and extrapolating that across enterprises makes the addressable pool large enough for several companies to reach $1 billion ARR each.

The broader portfolio read

Motamedi sees the next infrastructure wave centering on agent-specific abstractions: new data structures, databases, and runtimes built around long-running agent workflows rather than request-response patterns. On the physical side, Greylock's earlier autonomy bets include Aurora (which he describes as the largest standalone public self-driving company) and Nuro. He expects robotics to follow a similar trajectory as underlying model capabilities and training datasets mature.

The $36 million SciLake check is the clearest signal of how Greylock intends to deploy Fund XVIII: backing founders with a proven ability to build category-defining companies, at a check size calibrated to that track record rather than to the company's age.