Contrary Research Rundown #89

Compounding startups compounding in each other's directions, plus new memos on Perplexity, MasterClass, and more

Research Rundown

In 2021, Parker Conrad, the CEO of Rippling, coined the phrase “compound startup.” This idea is defined as a company that “addresses a range of point solution systems all at once to build a broad portfolio of interoperable products.” Some of the characteristics of a compound startup include: (1) longer / deeper R&D investment phase followed by better R&D efficiency, (2) larger overall TAM, and (3) better CAC payback and S&M efficiency. In effect, the idea is that if you invest deeply in building a product you will continue to expand your TAM. And as you penetrate that TAM, you’ll have better unit economics because each incremental product is easier to sell.

This week another company continued to demonstrate its compound startup characteristics when it launched its further expansion into yet another product category: Ramp*. The corporate card company has expanded dramatically since it launched in 2019. Imagined first as a corporate card focused on helping companies spend less, the product has since expanded into expense management software, bill pay, procurement, contract parsing, pricing intelligence, and vendor management.

Ramp had previously also enabled users to set limitations for travel spending. But recently, Ramp announced Ramp Travel, which allows users to book corporate travel right in the Ramp platform without having to go anywhere else. This expansion takes Ramp into yet another competitive landscape against tools like AmEx Global Business Travel, TravelPerk, and Navan.

Returning to the compound startup question, its clear that Ramp fits that same bill. The intense R&D investment has allowed the company to expand broadly to compete in several disparate categories. In not only addressing corporate cards, but a broad set of other functional areas, it expands Ramp’s TAM. The key remaining question is does this improve Ramp’s unit economics? According to a piece by Packy McCormick, it has.

From March 2022 to May 2024, total payment volume (TPV) has increased to $35 billion. Off of that payment volume, Ramp has grown the percentage of contribution it is able to capture. As Packy explains, “contribution profit represents interchange on card transactions minus the costs it incurs to process that volume, plus Bill Pay fees and Ramp Plus ARR (minus their costs, too).”

So Ramp has been able to invest heavily in R&D and expand its TAM, all while actually making more money on each transaction. The more products Ramp is able to sell, the higher the total profit margin becomes. And it is, in line with a compound startup, easier to sell each of those modules. Ramp is already installed as the corporate card for its customers; convincing a customer to expand what they can do with Ramp’s product isn’t nearly as expensive as initially landing the customer.

Going forward, more and more companies will vie to be the operational center of gravity for their customers finances and operations. Rippling has expanded to compete with Brex and Ramp. Ramp has expanded to compete with Navan and TravelPerk. Deel, Remote, Gusto—every company wants to own more of their customer’s universe. The question will just be who can most effectively compound to success?

*Contrary is an investor in Ramp through one or more affiliates

Perplexity, also known as Perplexity AI, is an AI-powered, conversational search engine that answers queries using natural language predictive text utilizing sources from around the internet. To learn more, read our full memo here and check out some open roles below:

• Machine Learning Engineer - San Francisco Bay Area

• Software Engineer - Full Stack - San Francisco Bay Area

MasterClass offers an online library of structured lessons from celebrities, industry leaders, and cultural icons, capitalizing on demand for unique online learning experiences. To learn more, read our full memo here and check out some open roles below:

• Staff Generative AI / ML Engineer - San Francisco, CA, United States, Waterloo, Ontario, Canada

• Product Marketing Manager, Consumer - San Francisco, CA

Check out some standout roles from this week.

• Flagship | San Francisco - Senior+ Product Designer

• Gretel | U.S. Remote - Staff Software Engineer, Backend Cloud Platforms, Staff Software Engineer, Backend, Machine Learning Solutions Engineer

• Harvey | San Francisco - Software Engineer, Frontend, Software Engineer, Backend Infrastructure, Software Engineer, Product

• According to a Stanford study on the reliability of AI legal research tools, most AIs have hallucinations rates of 17-33%.

• OpenAI released a method to identify how its AI models, like ChatGPT, represent certain concepts internally, which could help "dial down" unwanted behaviors and tune the models to favor specific topics or ideas.

• Leopold Aschenbrenner believes that the development of artificial superintelligence (ASI) by 2030 is inevitable, and that the only way to prevent China from using ASI to achieve global domination is for the US and its allies to get there first, even if it requires drastic measures like suspending environmental protections and placing all AI research under military-level security.

• Current and former OpenAI employees have issued a public letter warning that the company and its rivals are building powerful AI systems without sufficient oversight or ethical safeguards, and are “muzzling” employees who might expose irresponsible activities.

• Shutterstock's AI-licensing business generated $104 million in revenue last year, highlighting the growing demand for visual media to train AI models.

• Databricks announced it’s acquisition of Tabular, a data manegment company, for over $1 billion. The acquisition follows that of Arcion Labs for $100 million and MosaicML for $1.3 billion and Okera in 2023, to expand its capabilities in data management, interoperability, AI, and security.

• Billionaire VCs like Vinod Khosla, Reid Hoffman, and Marc Andreessen are locked in a battle over the future of AI, with Khosla and Hoffman advocating for regulation to mitigate risks, while Andreessen and his allies push for unfettered open-source development.

• Chatbots are evolving beyond simple text-based interactions, with startups and tech giants like Anthropic exploring "AI agents" that can perform tasks and adapt to new scenarios like playing video games.

• In a surprising move, Anthropic’s policy lead is advocating against AI regulation. “Policy is a ratchet... That means whatever power we assign governments today represents the floor of their power in the future…for this reason, I've found myself increasingly at odds with some of the ideas being thrown around in Al policy circles”

• Monzo, the London-based challenger bank, reported impressive growth (i.e. 2.5x revenue) and profitability, nearly $20 million. “Customers have an average of 38 friends who also use Monzo.”

• Shein, the fast-fashion e-commerce company, plans to file for a £50 billion IPO, which would be London's largest IPO ever.

• Linear created opinionated software — which is purpose-built for specific use cases, guiding users towards a default process — to address the problems they faced with legacy tools at previous high-growth startups like Coinbase and Airbnb. It can provide a more streamlined and effective workflow for teams, rather than letting chaos emerge from overly flexible tools.

• Ticketmaster and Santander both confirmed they had suffered a data breach by the same group of hackers who claimed to be selling customer records online.

• Addepar, a financial software and data platform, has grown its AUM to nearly $6 trillion as of April 2024, while its shares are trading at a discount to its Series F valuation in the secondary markets.

• Internet publishers are generating revenue by selling their content data to developers of large language models, which could potentially replace advertising revenue for some publishers.

• Jensen Huang says the future of AI lies in physical AI systems that can understand and operate in the real world and robotics will become pervasive across industries. “In the future, robotics is a much more pervasive idea…Everything is gonna be robotic. All of the factories will be robotic…Robots interacting with robots, building products that are robotic."

• In 2021, there were seventeen publicly traded enterprise software companies projected to grow >40%. By the end of 2023, there were zero companies in that growth bucket and only two companies projected to grow >30%. At the end of May 2024, there were zero.

• According to an estimate from Sequoia, “the AI industry spent $50 billion on chips from Nvidia to train AI in 2023, but brought in only $3 billion in revenue”

At Contrary Research, our vision is to become the best starting place to understand private tech companies. We can't do it alone, nor would we want to. We focus on bringing together a variety of different perspectives.

That's why we're opening applications for our Research Fellowship. In the past, we've worked with software engineers, product managers, investors, and more. If you're interested in researching and writing about tech companies, apply here!