In yesterday’s post (see here), I shared a quarterly letter written by an institutional investor that gives his thoughts on investing in public software companies. The letter references a 1997 email exchange between Warren Buffett and Microsoft executive Jeff Raikes, which caught my attention. Here’s that section of the letter:

SaaS is an amazing business model, which makes us reluctant to give up on the category entirely without deeper analysis. Buffett always says that the best business is a royalty on another fellow’s sales—someone else puts up the capital and takes the risk, leaving you a high-margin, capital-light, recurring revenue stream (particularly relevant during troubled times like recently with the Iran war, when traditional companies face oil/margin/ consumer demand risk and can become difficult to analyze). Traditionally, these royalty-like businesses were rare, expensive, and often no longer fast-growing. Software is a pure expression of that thesis (Jeff Raikes at Microsoft made this point to Buffett in his famous 1997 email) and happens to have the added benefit of continued high growth. When that royalty is offered cheaply, it’s worth looking into more deeply.

This “famous” email is new to me. I’m really curious to understand why it’s so famous and to learn more about Raikes’s and Buffett’s thinking on royalty-like businesses.

I’m going to see if I can dig up this email exchange. If I find it, I’ll share what I learn in another post.

‍

This weekend I was doing research on X about public software companies. I wanted to understand the lens other investors are looking through to view these companies in the age of AI. I found some interesting posts, and one from Tim Liu, founder of Meditation Capital Management, particularly caught my attention. He linked to his fund’s Q2 letter, which focused exclusively on the lens he uses for investing in (and avoiding) software companies. See the post here and the link to the letter here.

He makes some good points, and his framework for evaluating software was interesting. I noticed that software stocks are trading as a “basket” (i.e., they’re correlated and they move together), but I found that the stocks in that basket are very different. The companies all sell software, but they serve different types of customers (enterprise vs. SMBs) and solve different problems (e.g., marketing vs. financial reporting). I think this basket approach the stock market is taking to valuing software companies will provide an opportunity to savvy investors who do the work to understand which companies in these baskets are unique and will thrive in the age of AI instead of being displaced by AI.

The part I found most useful was his framework for thinking about how AI will impact software. A lot of his thinking is logical, but I disagree with his view that customers will rebuild and customize core software that’s key to their business operations. Some will, but the majority, especially SMBs, won’t. As I shared in this post, rebuilding and maintaining a critical system is a heavy lift that carries a ton of risk. Many companies can get a better return on the time, energy, and cost required to build a system from scratch. I think a more likely path is that companies will create apps to handle niche use cases (like specific processes) and integrate those apps into their off-the-shelf, mission-critical systems (i.e., pump the data back into the system of record). This will allow them to use their domain expertise to solve the problem in the way they see fit while avoiding a mammoth allocation of resources to build and maintain a new system.

Overall, I enjoyed Tim’s letter, and it’s a great read for anyone curious about how institutional public market investors are evaluating investment in SaaS companies in the age of AI.

I had a debate with an entrepreneur this past weekend about AI and software companies. The question was whether AI will disrupt mission-critical software companies. Think ERP, CRM, and HCM software like NetSuite, HubSpot, Salesforce, Workday, etc.

Having built an ERP system with CRM functionality, my answer is no. In the short to medium term, these software companies will continue to have a strong moat. I believe this for two reasons. First, these systems are very complex and run functions that are mission critical. The risk of replacing one of them with a system that doesn’t work as well is too high, even if the upside is saving money. Disrupted operations can lead to significant financial losses and tarnish a company’s brand. Most companies don’t want to take those kinds of risks (start-ups might, though).

When I was running my company, there was zero chance you could get me to change from the ERP/CRM system we built for ourselves. The risk and learning curve associated with switching were too high. Even if someone had given me the software for free, I would have said “no thanks.”

Cost is the second reason I believe mission-critical companies aren’t about to be disrupted by AI. Having AI build a system as complex as the ones mentioned above would take significant time and energy and cost a ton via tokens. Then there’s maintenance. You can’t just build it and forget about it; you have to maintain homegrown systems, which can require material resources. When a company thinks about the time, energy, and cost required to build and replace a system, they’ll keep what they have and allocate those resources to high-return activities.

My company’s internal software was a living thing. We were always making improvements and changes to it. I learned over the years to budget a certain amount of salary and team bandwidth for maintenance of this software.

Complex, mission-critical software is the backbone of many companies. If one of these systems stops working, a company is flying blind and in some cases can’t operate at all.

As of today, I don’t think these companies are at risk of losing customers. Whether they’ll continue to grow at historical rates is a legitimate question. I think the probability that they will is high, because I doubt that company leaders want to start building these types of systems from scratch. The return on the allocation of resources doesn’t make sense.

Yesterday, I shared (see here) that SpaceX was raising an additional $20  billion by issuing bonds. The deal priced today, and per Bloomberg (see here), the final order book has “less than three times the amount of debt offered . . . .” I took that to mean the demand was two or three times greater than the offer so orders totaled somewhere between $50 and $75 billion.

I don’t know a ton about bond issuance or bonds in general. I’m curious, though, and hope to get time to learn more about them. Public technology companies have issued a lot of bonds in 2026, and SpaceX just added to that list.

Last week, SpaceX completed the largest IPO raise in history (see more here): roughly $75 billion, it’s reported (see here). Now, the company is raising an additional $20 billion through bond issuance. The plan is said to be for the bonds to mature in between five and thirty years. It appears that the majority of these funds will be used to refinance a bridge loan that’s roughly the same size as the bond offering. They’re looking to convert the loan from high-cost short-term debt to lower-cost longer-term debt.

In the last month or so, we’ve seen several large tech companies raise tons of capital. And debt raises have been a material part of their capital-raising plans. I’m really curious to see how this plays out. I’m wondering if this is a window that companies are taking advantage of before it closes or, conversely, it will become the new norm for tech companies looking to raise significant capital.

SpaceX completed its IPO offering. According to Bloomberg (see here), SpaceX has raised a staggering $75 billion from investors by selling 555.6 million shares at $135 each. At that price, the company will have a market cap of roughly $1.77 trillion. For context, the second-largest amount ever raised during an IPO was $29.4 billion; that was Saudi Aramco’s IPO in 2019. SpaceX will begin trading on the stock market later today. I’m curious to see how receptive public-market investors will be to the company in the coming weeks and months. Regardless, this is a record-setting IPO.

Last week, I shared that Google raised $85 billion by selling stock (see here). That’s a lot of money, and it’s also the first time since 2006 that they’ve raised capital by selling shares. Today, a friend shared a Bloomberg article with me (see here). The article reports that Anthropic obtained a $35 billion loan associated with leases at five data centers. It goes on to say that Google agreed to backstop the lease payments at these data centers should Anthropic be unable to pay, but only after the data centers are fully operational.

The devil is in the details, and this deal is likely very complex. I don’t know anything about this space or have any unique insights. But this seems like a material commitment. Google has been flexing its financial muscles lately. I’m really curious to watch how they deploy capital and what the return (and risks) are on deals like this one.

Yesterday, I wrote about Google’s CEO presenting on the company’s plans to proceed with selling $80 billion in new shares. I’m a huge user of YouTube and its premium subscription, and it’s one of the apps I get the most value from every month. I get a great return on my $15.99. I dug into the detailed presentation, and one stat really stuck with me.

The CEO said that YouTube did $60 billion in revenue last year (page 11). I knew that YouTube is a big business, but that’s an enormous business. That number stopped me in my tracks. It got me thinking. YouTube is big enough to be a large, likely $100 billion plus, market-cap company on its own.

YouTube is a giant inside another giant. Interesting data point.

Earlier this week, I shared that Google’s parent company is raising $80 billion by selling stock (see here). And I wondered what it will do with all that money (it already had almost $128 billion in the bank before this raise). Well, the CEO has answered that question. Today, he shared the presentation he and the CFO gave to investors interested in participating in the raise. It seems they’re going to deploy it across the business to maintain or obtain a leadership position. So, not one single thing, but doing more of what they’ve already been doing.

If you’re interested in the press release about the presentation, it’s here. And for the full slide deck presentation, look here.

Today I read the press release from Google’s parent company about raising $80 billion (see here). As of the writing of this post, no company has ever raised that much capital in a single deal. For context, before this announcement, Google hadn’t sold equity to raise cash since 2006, when it raised $2.1 billion in an equity offering. And per its latest quarterly financials, it had almost $128 billion in cash on its balance sheet. So an equity raise, especially one this large, is unusual for the company.

Here’s how the $80 billion is structured:

• $10 billion – Berkshire Hathaway is anchoring the deal by buying via a private placement at a reported 6.5% discount to the previous day’s closing price.
• $15 billion – Issuance of mandatory convertible preferred stock, which, as I understand it, is a hybrid instrument. It’s a stock that pays a fixed dividend yield but converts to common stock on May 15, 2029. This stock will be priced this week.
• $15 billion – Marketed sale of common stock. Bankers will pitch this to institutions in the hope that they’ll commit to buying the newly created common shares.
• $40 billion – At-the-market offering, which means that shares of common stock will be periodically sold, likely on the open market. The company doesn’t have to pre-announce these stock sales; it can offer them as it sees fit. It’s reported that at-the-market sales won’t start until the third quarter.

This is a lot of cash and a big deal. I’m curious to see how all the pieces of the deal come together. And I’m even more curious to see how the company plans to deploy such a sum.