What's 🔥 in Enterprise IT/VC #350

AI in the enterprise - "tryers" vs. "buyers" - still early but...

AI is 🔥 and is the future, but just a reminder it’s still super early in the enterprise. I can’t tell you how many pitches we have with founders telling us about the enterprise traction from their AI product, but when you peak beneath the covers and ask about retention or what’s in production, the reality is much different. That’s 👌🏼 as early markets tend to be filled with “tryers” and not “buyers”. It makes sense as every Global 2000 board is asking its IT leadership about its AI strategy which has compelled many an organization to pilot some easy wins to show off AI. However, the gap between pilot and full enterprise roll out is still going to take time as many of the largest enterprises are still concerned about hallucinations, security and privacy…which means lots of opportunities for founders and investors.

“We at JPMorgan Chase will not roll out generative AI until we can mitigate all of the risks,” said Feintsmith during the first keynote address at Databricks Data + AI Summit 2023.

“You want to talk about responsible AI, you want to talk about hallucinations, you want to talk about misuse, you want to talk about having the right cyber capabilities so that the models aren’t poisoned or tampered.

Here’s Nasdaq’s CTO recently talking about all of the cool things they are building and hacking on but not yet in production.

Employees must register to use certain AI tools internally and undertake training before they proceed, Peterson said. Engineers so far have filed more requests for permission than legal counsel have, a spokesperson added. The company would not immediately ban tools like others, though.

"We're not going to go dark early," Peterson said.

Still, despite using other forms of AI for years, Nasdaq's latest work remains experimental; no code has been published yet drafted by AI, Peterson said. The company's lawyers are hashing out with vendors who owns the final output, he said.

This reminds me of the last AI hype wave from 2016.

Despite the parallels to 2016, we will get there much faster this time. Why? Because all of the investment made in the last wave has allowed enterprises to have the expertise and data infrastructure in place to take advantage of LLMs. Zooming back to today, here’s a chart from the latest Morgan Stanley CIO Survey with 14% of CIOs saying they are starting pilots but only 4% in production. This is the gap founders love to see as it means an opportunity to build. How quickly this gap closes will be interesting to see in future CIO surveys.

Enterprises will also get a massive assist from the large SIs and consulting firms as in the last few months, they’ve collectively announced $8B of spend on AI to help the JP Morgans and Nasdaqs of the world accelerate the adoption curve. Just this past week KPMG announced it will spend over $2B in AI and Cloud services while Wipro will spend $1b to train its entire staff in AI. Previously Accenture announced it would spend $3B on AI, Deloitte $1B on AI, and PWC $1B spend on Generative AI. That’s a shit ton of 💰 to be spent on AI which will help unleash a torrent of enterprise applications into production in the next 3 years. Exciting times are certainly ahead 📈!

As always, 🙏🏼 for reading and please share with your friends and colleagues.

---

Scaling Startups

1. 💯

   

2. 👇🏼this is what I call founders seeing around the corner - have to make sure though you don’t run out of runway before the next big step function and keep building…Superhuman is a great case in point - here it is back in 2014 before we invested and now - no huge data science and ML engineering team needed

   

---

Enterprise Tech

1. For infra founders - this is the chart from the Morgan Stanley CIO Survey! Only 29% of application workloads are in the public cloud today and it’s estimated to grow >50% in the next 2.5 years to 45% deployed 🤯.

   We’re talking $50B of spend on cloud with a second order effect of tens of billions 💰 that will need to be spent developing applications and deploying, monitoring, and securing them. The future is certainly bright although it doesn’t feel that way at the moment as many of the large public enterprise software cos continue to decelerate revenue from the COVID tailwinds

   

2. 🤯 Morgan Stanley Research projecting $40B of Microsoft AI Revenue…and a path to $3 Trillion Market Cap

   

3. Langchain has been amazing for developers to tie together AI workflows for prototyping but there is also a downside

   

4. Hugging Face to raise at $4B valuation (Forbes) - Github for ML

   Just over a year ago, Hugging Face raised $100 million in a Series C round led by Lux Capital; Coatue and Sequoia were new investors in that round, joining A.Capital Ventures and Addition. The company had attained a $2 billion valuation in that round despite taking in less than $10 million in revenue in 2021. Its revenue run rate has spiked this year and now sits at around $30 million to $50 million, three sources said — with one noting that it had more that tripled compared to the start of the year…

   Hugging Face makes money by charging for security and corporate tools on top of a hub of hundreds of thousands of models trained by its community of developers, including the popular Stable Diffusion model

5. The CSPM (cloud security posture management) wars continue with 🦄 Orca Security claiming that even bigger 🦄 Wiz stole trade secrets

   Orca’s lawsuit, filed in federal court in Delaware, claims that Wiz “was birthed from the very beginning as a counterfeit copy of Orca’s ideas.” It claims that Orca founder Avi Shua invented cloud monitoring software and says Wiz CEO Assaf Rappaport and his co-founders intentionally copied Orca’s product when they started their company in 2020, a year after Orca’s founding. Rappaport and his co-founders left Microsoft’s security business to found Wiz.

   Orca alleged Wiz repeatedly hired attorneys that were already working for Orca. The lawsuit, which hasn’t been previously reported, alleges that in 2021, as Orca was first trying to file its patents, it ditched an external patent attorney after learning the person had started working with Wiz to file patents for “overlapping” technology. Orca in 2020 also fired its corporate counsel, who had been attending board meetings, because the person was simultaneously working as a corporate counsel for Wiz “to assist Wiz in its attempts to copy Orca,” the suit alleges.

6. Great interview on developer_led with Zach Holman, engineer #2 at Github and also advisor to Gitlab on early days for both cos. This is something I preach and look for in founders - I ❤️ platforms but need to get hyper-focused with initial product and community.

   GitHub's founders had a deep understanding of the Ruby community, which allowed them to build a strong initial user base. By capturing the attention of influential developers and providing a valuable platform, GitHub started gaining traction organically.

   This is the same for Snyk - developer friendly security was not just for open source but more specifically the Node.js community.

7. From a couple of weeks ago from James Governor, RedMonk, on need for Golden Path when it comes to platform engineering and software development - guardrails needed

   That’s what we’re seeing. And I think that there was a bit of a myth over the last few years that, oh, you know, just let developers choose whatever they want. In some cases it did happen. And what you ended up with was a bit of a mess where, sure, the developers had built some great system, perhaps using open source components, cloud services, but then they ended up spending their time maintaining that system as opposed to building new application functionality. So what we want to do is we want to think about making life easier. Golden Paths. Some of the core advantages that we’re going to consider would be better developer onboarding. Too often today, that’s a really gnarly thing. Too many steps, trying to just get the laptop, get the environments installed. It can take a long time and that’s a productivity problem.

8. Easily forgotten when it just works

   

9. AI Ain’t Free - how do you price for AI compute? Salesforce takes first shot across bow - Salesforce Makes Rare Price Hike After Launching AI Features

   *List prices will go up about 9% across major products

   *Adding AI features are expensive due to computing required

   Many tech companies are trying to figure out how to profit from new generative AI features, which can be costly to offer due to the large computing resources required. In addition to the list price increases, Salesforce will charge extra fees for new generative features — last month it unveiled an AI “starter pack” that will cost $360,000 per year for 50 users.

10. To that end, so many pricing model questions with great breakdown here from two of my faves, Mostly Metrics interviewing pricing guru

    Kyle Poyar
    from OpenView

    How does AI actually make money?