WEKA Took Five Years to Ship, Then Hit $100M ARR and a $1.6B Valuation: Liran Zvibel's Story

October 5, 2026

TL;DR: WEKA is a software data platform that feeds data to GPUs fast enough that they stop sitting idle. It was founded in 2013 by Liran Zvibel, Omri Palmon and Maor Ben-Dayan, three members of the core team behind storage startup XIV, which IBM bought in 2008. Zvibel's first company after IBM, a consumer music-video app, raised tens of millions and failed. WEKA went the other way: deep tech, sold to large enterprises, and it took about five years to get a commercially ready product. In that time it raised a $10M Series A and a Series B with essentially no revenue, much of it from strategic investors like Nvidia, Qualcomm and Seagate. Once it found its use case in AI training, it doubled every year from 2020, and at the time of the interview in 2024 it had crossed nine-figure ARR. In May 2024 it raised a $140M Series E at a $1.6B valuation. Zvibel told the whole story on The Product Market Fit Show.

What does WEKA do?

WEKA makes data infrastructure software. Companies run it on standard servers with fast networking and flash drives, or on cloud instances, and it pools those machines into one very fast, very large file system. It replaces the storage product a company was running before. It does not sit on top of it.

The plain-English version Pablo used on the show: without WEKA, a company's GPUs spend most of their time waiting for data. With WEKA, they spend most of their time working.

Key stat: as Zvibel shared on the show, customers' GPUs typically ran at about 30% utilization before WEKA and 80 to 90% after it.

"It almost doesn't matter how much you're paying for the WEKA. And by the way, we cost exactly like the other product, but you're getting three times on your $30 million spend for compute." — Liran Zvibel, WEKA

Zvibel describes it as close to an operating system, with its own networking stack, IO stack and scheduler. The same software runs on-premises and in the cloud.

The bet was timing. In 2013, three things were changing at once: containers and microservices, a new standard for connecting flash drives to CPUs called NVMe, and a roughly 100x jump in networking speed over a few years. When WEKA started, Zvibel said, NVMe "was a PDF." The team bet that standard would win, and built on all three shifts.

The market they walked into was big and broken. Zvibel described data storage on the show as about $150 billion a year, split across hundreds of products, most of them software locked inside a vendor's own hardware box.

In June 2025, WEKA launched NeuralMesh, a storage system built from microservices for AI and agentic AI workloads, and in November 2025 it announced an "Augmented Memory Grid" extension on NeuralMesh, validated on Oracle Cloud Infrastructure, as verified via an October 2026 web search of the company's press releases.

Who founded WEKA?

WEKA was founded in 2013 by Liran Zvibel (CEO), Omri Palmon and Maor Ben-Dayan, as verified via an October 2026 web search. All three were part of the core team at XIV, a storage startup acquired by IBM in 2008. The company was founded in Tel Aviv and is now headquartered in Campbell, California, with an office in Tel Aviv, as of October 2026.

WEKA was not Zvibel's first try after IBM. In 2011, the team left IBM and started a consumer company, called Fusic in the episode description. It let teenagers record themselves singing and dancing alongside YouTube music videos.

"When we left IBM, we left it with a huge sense of hubris, basically. Anything we would do is going to turn out being an incredible success." — Liran Zvibel, WEKA

The music labels loved it. Zvibel said artists like Justin Bieber and One Direction ran contests on it, bringing millions of visitors and tens of thousands of recordings. But most teens lacked front-facing cameras and fast internet, so recording took real effort. Once a contest ended, nobody came back.

The lesson he took was to go back to what the team was best at: deep tech, sold to enterprises.

How much has WEKA raised?

All rounds below were verified via an October 2026 web search of WEKA press releases and contemporary coverage (TechCrunch, Calcalist, FinSMEs, StorageNewsletter):

$10M Series A, January 2014.
$22.3M Series B, 2016, led by Walden Riverwood Partners with Qualcomm Ventures. (On the show, Pablo rounded this to $25 million.)
$31.7M Series C, May 2019, from existing investors plus strategics including Hewlett Packard Enterprise, Mellanox, Nvidia, Seagate and Western Digital. That brought total funding to $66.7M.
$73M Series D, January 2022, led by Hitachi Ventures, later extended to $135M at a $750M valuation in November 2022.
$140M Series E, May 2024, led by Valor Equity Partners at a $1.6B post-money valuation, with Nvidia, Atreides Management, Generation Investment Management, Qualcomm Ventures and Hitachi Ventures among others. The round included about $40M of secondary for roughly 200 employees.

Total funding figures vary by source, from roughly $340M to $415M (the figure Pablo cited on the show). No new round had been announced as of October 2026.

Key stat: WEKA's valuation more than doubled in 18 months, from $750M in November 2022 to $1.6B in May 2024.

Five years with no revenue: how WEKA validated demand and stayed funded

Pablo's main question for Zvibel was simple. In deep tech you cannot build an MVP in a few months. So how do you know anyone will buy before you have spent years and tens of millions building it?

Zvibel's answer was that "can we build it" is not the biggest risk. The biggest risk is building something hard that solves a pain nobody cares enough about.

"Buying from a no-name company is a huge risk. You may not be there and all of these buyers have made a mistake of picking something that looks like a cool technology from a company that didn't make it, only to go and revert to the tried and true big players that at least you know are gonna be there next year." — Liran Zvibel, WEKA

So WEKA ran proofs of concept early, before the product was finished, to find a pain that was big and repeatable. The first segments did not fit:

Hedge funds cared about scale, performance and latency, but there were not enough of them to build a big company.
Life sciences could have worked, but customers were conservative and slow to convince.
Nuance, the speech recognition company, was one of the first places the team saw GPUs solving an enterprise problem, "before AI was cool." It ran an early alpha with them.

Then in 2017 and 2018, AI training took off, and every team doing it hit the two problems WEKA was best at: performance and scale.

Staying alive through those years took money with no revenue to show for it. Zvibel used two moves. First, the POCs gave him excited customers to point to, even without purchase orders, and he raised from a deep tech investor that knew how to interview those customers. Second, he raised from strategic investors, a lot of them.

"We've actually funded ourselves through a lot of the strategic players. And now some of them are competitors, but throughout the run B and C, we raised money from Qualcomm and Melanox and Nvidia and Micron and Seagate and Western Digital and Hewlett -Packard and Cisco and Hitachi." — Liran Zvibel, WEKA

Founders often avoid strategics because of signaling risk: if your one strategic backer loses interest, what does that say? Zvibel's fix was to have many. Each wanted a look at the future: server makers wanted to sell the hardware WEKA ran on, and component and networking makers wanted to see what would be needed next. That broad backing then helped with VCs and early customers.

Never miss a founder's PMF story

Subscribe to The PMF Show

"And end of the day, if you're running a deep tech company and you have a strong belief that what you're doing is differentiated and you're gonna hit product market fit, the most important thing is don't run out of money." — Liran Zvibel, WEKA

The product market fit moment: beating "free"

WEKA's first real proof came in the second half of 2017, with what Zvibel called one of the most ambitious full self-driving projects in the Bay Area. That team was running on Pure Storage, which at the time marketed itself as the best storage for AI.

Key stat: as Zvibel shared on the show, WEKA cut that customer's "time to epoch," one full training cycle, from about 14 days to four hours. A week on WEKA produced more output than a year on the incumbent.

The incumbents responded with a classic move: they offered their product for free. Zvibel said big vendors use this to suffocate new entrants before they land their first dozen customers.

"And basically when they were buying us instead of taking the free, we knew we've had it." — Liran Zvibel, WEKA

That was the first of two PMF moments, at the end of 2017 and start of 2018. The second came in 2020, when WEKA built a real sales team. New salespeople, who were not founders and had not been around long, were closing six- and seven-figure deals with strangers.

On pricing, WEKA tried being clever and then stopped. It tested charging for performance gains and for the number of servers. Each made the comparison to incumbents too hard for buyers. It settled on what the industry already did: an annual subscription based on capacity.

"We've decided reducing friction is more important than charging for the absolutely right parameter. And we're just charging like the rest of the industry." — Liran Zvibel, WEKA

The reason is procurement. Once an incumbent sees it might lose a deal, it will fight, so the gap between "the champion wants to buy" and "we have a PO" must be short. A familiar model makes that easy: whatever you would have paid Pure, EMC or NetApp, pay WEKA instead.

Key stat: as Zvibel shared on the show, WEKA likes entry deals of about $200K to $250K ARR. Against a three-year appliance purchase that is $600K to $750K in software, and with hardware the customer spends $1M to $1.5M in total. WEKA also has deals of seven and eight figures in ARR.

WEKA kept 2018 and 2019 to a handful of customers on purpose, focused on value rather than revenue. It started scaling in 2020, and Zvibel said it doubled every year from then to the 2024 interview, reaching nine-figure ARR and beating its own plan.

Key lessons from Liran Zvibel's playbook

1. Play to your strengths. A team that had built and scaled enterprise storage tried consumer social and failed. It went back to deep tech for enterprises and built a unicorn. 2. In deep tech, the biggest risk is the pain, not the tech. Run POCs before the product is done. Look for a pain that is big, repeatable, and owned by enough customers to scale. 3. If an incumbent's free product loses to yours, you have PMF. Cutting 14 days to four hours made WEKA worth paying for when the alternative cost nothing. 4. Be boring on pricing. Charge the way the industry already charges. The faster procurement can say yes, the less time the incumbent has to fight back. 5. Get many strategics, not one. A crowd of strategic investors validates you to VCs and customers. One alone is a signaling risk. And when VCs cannot follow the story, the dollars still keep you alive.

His one piece of advice to his younger self: do not chase purchase orders before the product is ready. VCs push deep tech companies to sell early, and he says WEKA wasted resources in 2016 to 2018 doing it.

"Then, when you have something that you think is an MLP, try to really sell it. Anything before is just going to mount frustration on both ends." — Liran Zvibel, WEKA

MLP is his "minimum lovable product": enough to remove the friction that stops a customer from buying.

For more founder stories about raising before revenue and selling to big companies, see Chainguard's $50M Sequoia raise with no revenue, Pinecone's path from 40 VC rejections to a $100M a16z round, and our guide to getting your first enterprise customer.

Listen to the full interview: His 1st startup failed, but his 2nd one hit $100M ARR & a $1.6B valuation.

FAQ: WEKA

Q: What is WEKA? A: WEKA is a software data platform for AI and other high-performance workloads. It runs on standard servers or cloud instances and feeds data to GPUs fast enough to keep them busy. Its current platform, launched in June 2025, is called NeuralMesh.

Q: Who is the CEO of WEKA? A: Liran Zvibel, who co-founded WEKA in 2013 with Omri Palmon and Maor Ben-Dayan. All three were on the core team of XIV, a storage startup IBM acquired in 2008.

Q: How much funding has WEKA raised? A: As of October 2026, roughly $340M to $415M depending on the source. Its most recent round was a $140M Series E in May 2024, led by Valor Equity Partners at a $1.6B valuation.

Q: Who are WEKA's investors? A: Valor Equity Partners, Hitachi Ventures, Walden Riverwood Partners, Generation Investment Management and Atreides Management, plus strategic investors including Nvidia, Qualcomm Ventures, Hewlett Packard Enterprise, Micron, Seagate, Western Digital and Mellanox.

Q: What is WEKA's ARR? A: At the time of the 2024 interview, Zvibel said WEKA had reached nine-figure ARR after doubling every year since 2020. That is his figure from the show; WEKA has not published audited revenue.

Sources: Listen to the Full Founder Story

Liran Zvibel, Co-Founder and CEO of WEKA — failed with a consumer app, then spent five years building deep tech for enterprises, funded it with strategic investors, and grew WEKA past $100M ARR and a $1.6B valuation.