🧠Is Palantir the SAP+Salesforce of the AI Era?🤔

Ben Thompson suggests Palantir is best positioned to be the key AI platform company for enterprises. I'm a big fan of Thompson's work but I found his arguments somewhat incomplete. Some further thoughts.

✅The Bull Case : Palantir's leadership deserves credit for their first principles thinking. They addressed a fundamental issue: that implementing effective analytics projects in large organisations is painful. And it's primarily an integration and data problem.

While competitors were content to exploit existing architectures and sell to internal IT teams, Palantir boldly built technology (integrations, visualisation& DevOps tooling) and provided services (hands-on field engineers) which other BI/analytics providers never considered as their scope.

This gave them more skin in the game and allowed a compelling narrative of quicker value delivery, especially in large, messy brownfield architectures. Despite much higher costs, this resonated with governments and large enterprises struggling with hundreds of transactional systems and competing off-the-shelf products.

Transversal analytics projects tend to have more senior sponsorship on the customer side compared to yet another transactional SaaS system. This made Palantir's narrative more effective.

There are similarities to what SAP did for Finance and Operations half a century ago, and what Salesforce did when enterprise software delivery over the internet was still uncertain. Build holistic solutions for previously disparate areas, offering quicker paths to value.

Now, Palantir promises to do the same for AI agents. These agents extract intends - from user inputs, insights, or others signal to then an action on a transactional system. There's a strong technical argument for a direct line from an insights platform to AIP, Palantir's AI agent platform.

This technical potential seems to be driving Palantir's impressive stock performance.

❎The Bear Case : I think is on the business and strategy side.

Palantir’s USP is fit for the large legacy organisations of the 20th century - large governments, industrials, manufacturing, Pharma etc with a lot of legacy architecture.

21st-century digital-native post internet companies have fewer legacy issues like this. Focused solutions like Databricks for data science or Snowflake for enterprise might fit better.

It's unclear also if Palantir's expensive model will work for SMEs with limited IT budgets.

An agent platform like AIP could potentially commoditise the transactional platforms it integrates with. This isn't an ideal position for established players like SAP or Salesforce, who may respond with different pricing models or more closed ecosystems in the long term.

🙏Conclusion: This is among the most interesting high-stakes area in enterprise software. Early trends suggest potential consolidation, and Palantir's execution in difficult use cases does position them as a likely key player.

#ai