Key points
- Somorjai is skeptical of AI-native CRMs like Lightfield and Monaco, saying too much funding chases the category.
- He predicts a venture reckoning, calling valuations unsustainable with 1% or fewer companies going public.
- Salesforce's Slack bet pays off: revenue nearly tripled since 2020 and Slack is now its fastest-growing product.
How Salesforce buys its way to growth
Salesforce has turned corporate development into a growth engine, spending close to $100bn across 29 acquisitions. Ideas come from customers and a ventures team that meets thousands of companies a year. AI engineers are scarce and building is slow, so buying is often faster, which is how Slack and Fin arrived. Somorjai expects the money pouring into AI-native startups to end badly for most of them.
Takeaways
- Acquisitions are Salesforce's growth engine. Somorjai has helped oversee close to $100bn of deals, 29 acquisitions and investments in more than 750 companies. Service Cloud, now almost $11bn in ARR, was built on the 2008 purchase of Instranet.
- Three tests decide every acquisition. Targets must fit strategically and serve customers, Salesforce must be able to accelerate the business and fit its culture, and the deal must create value. Valuation is the hardest part because everything is expensive.
- Buying beats building in AI. Hiring AI engineers is very hard and a typical acquisition takes six to nine months to close, so buying can be faster than building. Salesforce could not build outcome-priced AI fast enough on its own.
- Slack's thesis proved bigger than expected. Bought in 2020, Slack is now Salesforce's fastest-growing product, with revenue nearly three times the level at acquisition. Somorjai says the thesis holds because Slack is the perfect conversational interface for AI.
- Fin proved outcome pricing works. Fin resolves 65% of cases without a human and handles more than 40% of traffic on help.salesforce.com, with an average sixteen days to value. Immerman says enterprise AI vendors now take a larger share of customers' labour budgets.
- Ventures moved from ecosystem to frontier AI. Salesforce Ventures began in 2009 with ecosystem bets like DocuSign and nCino, later Snowflake and Wiz, and raised a $1bn AI fund in 2023 backing Anthropic, Mistral, Cohere and Eleven Labs. It now funds physical AI including Figure AI and Zipline.
- Somorjai expects a valuation reckoning. He says too much funding is going into AI-native CRMs like Lightfield and Monaco and many valuations are unsustainable, which will hurt employees too. Roughly 1% or fewer companies go public, and above $10bn the buyer universe is tiny.
What kills a deal
- Weak financial controls.
- ARR that masks customer attrition.
- Security failures, the one diligence function that can kill a deal. One penetration test reached a target's source code and the CEO's laptop, and the problems were unrecoverable.
- His advice to founders is to red team their own stack.