Intro to the Sales Wiz Review Series

This is a new series I am starting, where I will evaluate the hottest tech startups and rate how good of a selling environment they appear to provide.

My goal will be to use public information online to assess the company’s financial state, employee satisfaction, product positioning, market fit, and other key indicators that are important to a salesperson vetting a new company.

The hope is I can help salespeople as they evaluate startups as potential places to work, using my past years of experience selling at startups to give my thoughts.

For my first review, I will be looking at Hugging Face, a Series C machine learning startup founded out of New York.

For this analysis, I will be looking at Hugging Face’s financials, product, market / industry fit and then will give a forward looking opinion on if I would go become a seller at this company in its current state, as of early 2023.

Keep in mind that startups are dynamic, things can change rapidly, so take this review with a grain of salt depending on when you are reading it.

Financials and Funding

Hugging Face is quite well-funded. They raised their Series A round in December 2019, then have consistently raised subsequent rounds of funding every year.

Their Series C raise of $100M was completed in May 2022, towards the end of when liquidity was still relatively available. The 2nd half of 2022 was quiet for startup fundraising in general, so it’s good for Hugging Face that they raised earlier in the year.

The lead investor for Hugging Face’s Series C was Lux Capital, a fund with $4B assets under management (AUM) who invests with check sizes from $100K to $100M, according to their website.

Looks like the Hugging Face round would’ve been towards the upper end of their average check size.

Lux Capital is not necessarily one of the top VCs you might immediately think of, but it has been around since 2000 and has had some exits, most prominently companies like Zoox (autonomous vehicles), Rigetti Computing (quantum computing), and Science 37 (clinical research).

That being said, for having been around since 2000, I would say looking at their list of exits, they’ve had modest success when compared to VCs like Sequoia Capital, Andreeson Horowitz and others.

You may think this is a lot more detail than necessary into Hugging Face’s investment background, but I argue that the details are necessary.

Just like you report to a VP of Sales or a CRO, the CEO of your startup reports to the board of directors. Usually the lead investor, Lux Capital in this case, and potentially other investors will have a board seat and will have influence over advising the CEO and rest of the management team.

They can make or break a startup depending on the advice they give or strings that are attached to money that can be allocated for future rounds.

In Hugging Face’s case, Lux Capital does not have too much background or success in AI or machine learning specifically, so I am not sure how much of a value add they would be as investors.

Furthermore, their founder seems to be on a crusade in the defense tech space, so I’m not sure how much focus they are putting into the AI investments space at the moment.

That being said, Sequoia Capital, a prominent VC investor behind exits like NVIDIA, Github and Apple, did invest in Hugging Face’s Series C round. So either they got FOMO and jumped in, or there is something promising about Hugging Face’s trajectory and growth potential (or a little bit of both).

Revenue-wise, there is not much I can find about Hugging Face. It is worth noting that many AI/ML companies raise money off of hype and future potential, without the revenue expectations that might be present in other more mature industries like data visualization and analytics software or HR software.

The only somewhat up-to-date revenue numbers I could find was on this website called Growjo (not to be confused with GoJo, from our favorite show Succession). Growjo has estimated Hugging Face’s revenue at $24.7M, and we will assume this was their 2022 revenue (source). Regarding accuracy of Growjo’s data, I am not sure, but this is what I found online:

Since we are looking at high-growth startups that are not generally cash flow positive, we will look at the annual recurring revenue (ARR) multiple to see if the startup’s valuation is justified.

If we assume Hugging Face’s revenue is $25M, and its latest raise yielded a valuation of $2B, then we get a 320x multiple.

To give you a comparison, Dataiku, another player in the machine learning space which is still a private company has $150M in ARR at a valuation of $3.7B, giving them a multiple of around 25x. This gives you an idea of the lofty expectations investors have for Hugging Face.

If you look at the investment thesis from Sonya Huang at Sequoia Capital, who led the Hugging Face investment, there is a lot of talk about Hugging Face’s underlying transformers models based approach to machine learning and their community users loving them, but that is about it.

I don’t see necessarily the translation to the B2B space, or any talk about enterprise adoption. This investment thesis mostly seems to be focused on the community-based virality Hugging Face has had, and the fact that they are at the bleeding edge in the AI/ML industry.

Product and Market Fit

On Hugging Face’s website is where I start to see some logos from companies like Google and Microsoft. I assume most of the users of Hugging Face are data scientists and ML engineers at these companies.

Oh and they have 81,103 stars at the time of writing on their Github repository. Impressive, but not sure what that would mean to me as a seller at this company.

Going to their pricing page, it looks like Hugging Face has some combination of a free model plus a usage based model.

Honestly, this is pretty confusing for anyone who is not deep into the AI/ML industry. I suppose most users would start at HF Hub, then gauge which offering they need to upgrade based on usage in the free version.

Once I clicked on the enterprise plan, I did see a nice case study from one of Hugging Face’s customers talking about how they had used Hugging Face to build a production natural language processing pipeline to handle messages that come into the business.

The efficiency gain story of having 20% more messages automatically handled is great, and I’d want to read more of these type of case studies if I was an enterprise buyer.

Overall, the Hugging Face product seems to be in a very innovative space, even within an innovative industry (AI/ML), but I really am missing the translation of its core technology to business value.

Production ML models are mentioned a few times, but to me this still seems like a community, try for fun, R&D at best play. I am surprised and quite impressed if Hugging Face is actually doing $25M in ARR, despite this more community-driven product positioning!

Company Outlook
Disclaimer: This is not career or investment advice, please do not take it as such.

Looking at Hugging Face’s latest job postings, I see mostly sales and marketing roles that they are hiring for from AEs and SDRs to a head of customer success.

I would imagine Hugging Face is under pressure from investors to start living up to the lofty valuation they have been given. A community is great, as long as Hugging Face can help those free developers move to paid accounts eventually.

When I see a company that is still so developer oriented, hiring several sales roles, it worries me a bit.

I don’t see any messaging on their website yet to suggest that Hugging Face is ready to support sales teams in the form of robust product positioning for the enterprise, case studies and testimonials and integrations with enterprise tooling.

I would hesitate to join Hugging Face as an enterprise seller who wants to close deals and make good money off of commission.

Rather than hiring more sales resources, I think Hugging Face needs to start experimenting with some true enterprise product marketing and building out a qualified inbound lead funnels to start supporting a sales team.

This funnel may potentially exist already, but I could see lead quality not being great at the moment, because Hugging Face still looks like an almost B2C, virality driven community play on their social platforms and website.

The good news is that currently there are some self-serve plans, so that the potential enterprise leads can more easily funnel into the “Contact Sales” bucket. Developers who are not ready for enterprise level usage can choose the free or self-serve options.

However, the self-serve options are numerous and potentially a bit confusing for a non-technical user or buyer. I could see some prospects opting to talk to a sales person to get guidance on the various options they have.

Maybe there is an opportunity to consolidate the various plans into Free, Pro and Enterprise buckets to help non-technical buyers decide which plans their developer teams might need. Then go talk to sales if it makes sense.

It’s fine to not be ready for an enterprise sales team. If the PLG, self-serve route is working decently well, then maybe it’s worth doubling down on that core set of users.

This is the journey Docker went through. Initially trying the enterprise sales route, blowing through VC money, and missing their ARR targets. Then downsizing from 400 to 60 employees, and going self-serve which completely transformed the company.

If you are interested in this story and how Docker looks today from a GTM perspective, I will cover the company in a future newsletter.

A high potential company that needs to revisit product and marketing

In conclusion, Hugging Face’s technology has high potential, but to really grow their ARRs in a hockey stick fashion that mimics their Github stars count, I think they need to revisit their product positioning and translate the innovation they bring for enterprise CTOs and technology buyers into business value.

This could be done by doubling down on customer testimonials, case studies and integrations with enterprise tools the existing customer base already uses.

For now, I will keep watching Hugging Face, and how they continue to iterate on their developer-led go-to-market strategy (GTM).