MESSAGE
| DATE | 2025-12-25 |
| FROM | Ruben Safir
|
| SUBJECT | Subject: [Hangout - NYLXS] Open Source Project With Little Revenue In Talks
|
Open Source Project With Little Revenue In Talks To Raise At Least $160
Million
The startup behind popular Github project vLLM is out fundraising, as
venture capitalists hunt for companies building tech that can make AI
systems run more efficiently.
ByIain Martin,
Forbes Staff
andAnna Tong,
Forbes Staff.
Dec 22, 2025, 04:32pm EST
Intricate Network of Blue Cables in Modern Data Center Facility
Getty
Investors are about to wager hundreds of millions on a startup with very
little revenue, betting that AI is hot enough to turn an open-source
codebase into a thriving business.
Spun out of UC Berkeley from the lab of Databricks co-founder Ion
Stoica, vLLM has in recent weeks been pitching investors on raising
money to commercialize a popular open source codebase that speeds up
large language models, and squeezes better performance out of AI chips.
Four sources have independently told Forbes that vLLM’s co-leader Simon
Mo has been taking meetings around Silicon Valley looking to raise $60
million now and at least $100 million in a second tranche. Mo did not
respond to a request for comment.
The sources said the funding could give the startup a valuation of
around $1 billion. The numbers are in flux and could change, they cautioned.
For now, vLLM doesn’t appear to have a website or a business model to
monetize its code beyond donations. So far, it’s banked $300,000 to date
including checks from Sand Hill Road blue blood firm Sequoia.
The nascent company is one of a wave of new startups staking claims in a
newer, growing part of the AI economy. OpenAI, Anthropic and big tech
companies have spent tens of billions of dollars training new large
language models but as AI startups sign on customers, more and more
spending is focused on supporting the running of trained AI systems,
which is known as inference.
That’s become a more and more expensive proposition for AI startups as
they serve customer’s short text prompts, complete complex coding tasks
or generate images and videos. For example, OpenAI could be spending
more than a quarter of what it’s making on its Sora video creation app,
the economics of which Sora head Bill Peebles called “completely
unsustainable.”
That’s turning investors’ heads to tools like vLLM, that promise to keep
a lid on the cost of running AI processes. Fireworks raised a $250
million round in October on a $4 billion valuation in October for its
cloud-based tool, while its rival Basten closed a $150 million fund
raise on a $2.15 billion valuation a month earlier. Photo and video
inference specialist Fal raised $140 million on a $4.5 billion valuation
earlier in December.
vLLM’s pitch has grabbed the attention of venture capitalists despite
the fact that the startup is currently building community-driven,
open-source software “libraries” hosted on Github. That’s because its
software, which works by using a GPU’s memory more efficiently, allowing
it to distribute tasks across fewer servers, has emerged as one of the
most starred on the developer focused site. And unlike Fireworks,
Baseten or Fal, startups can use vLLM to optimize inference on their own
chips and servers.
Investors might be willing to look past vLLM’s lack of commercialization
so far because of the long track record of open source tools and
projects like Red Hat, GitLab, and MongoDB morphing into critical tech
tools with multi billion dollar valuations to match. Red Hat notably
scored one of the largest exits of all time for a software company when
it was snapped up for $34 billion by IBM in 2019, after growing from a
reseller of open source software tools into the developer of a
commercial operating system that generated $3.4 billion in revenue at
the time of the deal.
“There’s going to be hundreds of billions of dollars spent on
inference,” said Dylan Patel, founder of AI research group Semianalysis.
“Redhat makes so much money developing open source software and then
building on top of it with services for Linux. So yes, I think it’s
possible for vLLM to do the same.”
Rashi Shrivastava contributed reporting.
--
So many immigrant groups have swept through our town
that Brooklyn, like Atlantis, reaches mythological
proportions in the mind of the world - RI Safir 1998
http://www.mrbrklyn.com
DRM is THEFT - We are the STAKEHOLDERS - RI Safir 2002
http://www.nylxs.com - Leadership Development in Free Software
http://www.brooklyn-living.com
Being so tracked is for FARM ANIMALS and extermination camps,
but incompatible with living as a free human being. -RI Safir 2013
_______________________________________________
Hangout mailing list
Hangout-at-nylxs.com
http://lists.mrbrklyn.com/mailman/listinfo/hangout
|
|
