A Startup Is Trying to Buy PayPal… Craziest Deal of 2026! | E2312

Stripe is reportedly pursuing an acquisition of PayPal, with the deal reported on July 15, 2026, and valued at more than 50 billion dollars. PayPal's share price peaked above 300 dollars during the 2021 to 2022 era, and the company may go private at approximately 16.50 per share under the deal. Jeff Morris Jr. found the idea of PayPal being acquirable surprising given its iconic status, but noted Stripe's private structure gives it flexibility to take a longer-term view without public market pressure. Eric Bohn was less surprised, arguing PayPal had been stagnating for years as its culture calcified after MBAs joined and innovation stopped, and that PayPal's new CEO, installed around February, had already signaled large cuts were possible.

Serious structural problems in private markets were flagged, including opacity, lack of price discovery, and friction in secondary transactions. The USBC and Andreessen Horowitz dispute over Series H shares was cited as an example. Foreign investors using Singaporean shell companies have attempted to access US defense-related private company cap tables, and Andreessen Horowitz was described as exercising strict control over its equity to prevent unwanted investors. SPV managers have in some cases disappeared and become unresponsive, and a prediction was made that the SEC or a similar body will likely be required to clean up the SPV segment.

Fin, previously known as Intercom, pivoted toward AI agents, renamed itself after its agent product, and was acquired by Salesforce, described as a SaaS unicorn that struggled in the AI era but achieved a solid exit. Webflow was described as revolutionary during the Web 2.0 era but caught off guard by the AI era shift in design paradigms, leading to major staff cuts to recapture a startup culture, though whether it can catch up to competitors remains to be seen. FinTech broadly was characterized as having been slightly overlooked by investors over the last two to three years relative to pure AI companies.

Eric argued that Claude Opus and what he called Claude 5 are performing at a level comparable to senior engineer feedback, and that AGI has effectively been reached within the coding domain. Jeff added that model companies shifting focus toward physical AI signals they believe they are close to solving coding and digital-native use cases, and Jason predicted physical AI and robotics will reach sophistication comparable to modern self-driving within the next couple of years. Jeff noted startup clusters in physical AI form more slowly than in software because of hardware complexity and ties to traditional industries.

Very few pre-Series A portfolio companies are building their own evals, and the conversation about custom evals is recent and driven by the number of startups destroyed by OpenAI and Anthropic shipping competing features. Eric said at pre-seed stage companies use whatever model is cheapest and most available, and the custom evals problem only becomes real after product market fit when a company needs to build a moat. Jeff noted that even without ingesting data for training, tool call metadata and customer behavior patterns reveal directional business intelligence, citing Satya Nadella's observation that frequency of tool calls alone is useful information.

Jeff identified FinTech and banking as categories that cannot be fast-followed due to regulatory, trust, and security barriers, and highlighted Erebor as a company that obtained a US banking license faster than any company previously and has Palmer Luckey on its cap table. He said AI-native fintechs and anything combining money with AI will produce significant winners, including the concept of giving an AI agent a bank account to act as a personal financial manager, and said Robinhood is unlikely to win that space because its product stack is too busy. Robinhood has nonetheless made tokenized assets and on-chain equities a central part of its publicly stated roadmap, reflecting a broader industry shift toward bringing traditional financial assets onto blockchain rails.

Jeff argued that Anthropic's early public communications framed AI in scary terms around job loss, damaging public support, and that failing to build a positive narrative risks the United States falling behind with national security consequences. Anthropic and OpenAI now have large teams and major headquarters in Washington DC working with both parties on AI policy, and speakers warned that making AI a partisan issue would slow US progress and lead to policy ping-ponging between administrations. There are rumblings of an executive order on open source AI in the United States, and China is possibly restricting open weight models from release.

Eric said the majority of VCs have not meaningfully used tools like Claude Code or committed anything to GitHub or Vercel, and that he still sees pen and paper being used in meetings with no clear path to institutional knowledge. Jeff said every employee needs to burn the boats on what they think their job is going forward, and added that it is relatively easy to become AI native within a company simply by being the most interested person in AI among your peer group. DAOs were once treated as a legitimate venture-backable category but proved highly inefficient in practice, with roughly 200 vocal decision makers slowing execution, and large society-changing crypto concepts like DAOs and the broader ownership thesis have not materialized as originally anticipated.