AI = Software + Services; From product to sales CEO; The cost of hacking you; Data security beyond DLP, and more

AI = Software + Services?

AI companies are emerging as an entirely new class of business, with some elements that resemble traditional software and others that are more like services. Anecdotally, we see gross margins for AI companies around 50-60%, somewhere between software’s typical 60-80%+ and services’ 30-50%. Why? ML and deep learning demand more cloud compute, not to mention humans to curate and maintain data. Scaling is hard, because it requires uncoiling the long tail of edge cases. And technical differentiation (and defensibility) is more challenging, as data becomes more commoditized and more models are open source.

Even so, we remain optimistic about AI. While the challenges likely won’t be eliminated, they can be minimized or turned into strengths. Founders can embrace services to capture greater value from customers. They can reduce cloud costs by reducing model complexity and optimizing their ML tooling. And perhaps the most important thing a founder can do? Find the right problem domain.

Read more here:

• Read the full piece on “The New Business of AI”
• Twitter thread from Martin Casado
• Twitter thread from Anaconda co-founder and CEO
• Discussion thread from Hacker News
• Other commentary that asks if startups could shift cloud costs to hardware costs?

Martin Casado is a general partner at Andreessen Horowitz, where he focuses on enterprise investing.

• Follow
• X
• Linkedin

Matt Bornstein is a partner on the enterprise team at Andreessen Horowitz, where he focuses on new data systems and technologies underpinning artificial intelligence.

• Follow
• X
• Linkedin

From Technical to Product to Sales CEO

Many startup founders evolve from “technical” to “product” to “sales” and go-to-market at scale CEOs. But that journey isn’t as neat or as linear as it looks. In these two podcasts recorded last year, we go deep with founders/CEOs who went on those journeys – tackling questions such as: What to focus on when? How do you build the right team for the right play, and at the right time? And how to define product-market fit when it hinges on users understanding the business value of your product?

Listen/read here:

• What time is it? Hard-earned lessons learned (with David Ulevitch and Sonal Chokshi)
• What comes first? Product-market-sales-fit (with Jyoti Bansal, Peter Levine, Satish Talluri, and Sonal Chokshi)

a16z editorial

Data Security: Evolving Beyond DLP

When a new platform emerges, a new ecosystem springs up around it – but the current emerging enterprise data ecosystem is missing next generation data security. As more data and applications migrate to the cloud, endpoint devices proliferate, and APIs increase vulnerabilities, the existing toolchains (e.g. legacy DLP 2.0, CASP 2.0) are inadequate.

Organizations want and need a solution that tracks data throughout the entire lifecycle and allows for role-based controls. IT security budgets have grown 10% year over year as security goes from afterthought to critical line item. It’s not just about security, however; it’s also about compliance and data privacy. Since 2011, data-driven risks have already increased the cost of compliance by 45%.

Bottom line: the market is ready for tools that provide a unified view, can handle real-time data, and are easier to integrate.

Joel de la Garza is a partner focused on information security related investments and other chaos adjacent businesses.

• Follow
• Linkedin

From Our a16z Summit

You Are the New Attack Vector
From business email compromise to SIM ports, cyberattacks have shifted from networks to you. It’s not only easier to hack people than it is companies today – and that includes employees of organizations – it’s also incredibly profitable. Martin Casado follows the money – and the trends – in this talk.

The Open Source CIO
Peter Levine also chatted with Mike Kelly, CIO of Red Hat during a16z Summit about why there will never be another Red Hat, when an open source project becomes a product, and how founders should think about M&A beyond the spreadsheets. Listen here.

a16z editorial