September 9, 2024 - Some notes on Snowflake:

Snowflake's stock-based comp is very high. It exceeds operating income, net income and cash flow from operations. This is concerning because SBC is an actual transfer of value despite the company’s decision to add it back to adjusted earnings. SBC made up 55%, 30%, 51%, 51%, 43% and 43% of revenue between FY 2019-FY2024.

The company competes with companies like Amazon, Google, Meta, and Microsoft for talent, but these companies have much higher budgets at which they can pay employees. To compete with these companies for talent, Snowflake can’t afford to pay comparative salaries since they are a much younger company in terms of its growth cycle and haven’t scaled enough yet, so Snowflake must make up for the difference in compensation by paying with shares.

Snowflake has built a very effective data warehouse for analyzing and querying structured data.

Structured data is data that is organized and can be displayed in rows and columns, like a table. This data is for business analysis and the reports are typically used by finance departments, marketing departments and sales departments. It shows yesterday’s performance.

Snowflake is not effective at handling unstructured data.

Unstructured data is data that is not organized. Examples of unstructured data are text in a .docx file, .jpeg files, images, .mp3 or video files.

Databricks has excelled at analyzing and querying this kind of data. Databricks also excels at data lakes.

Unstructured data is stored in data lakes and structured data is stored in data warehouses but there has recently been a move to combine both, resulting in what is known as a lakehouse. A lakehouse attempts to combine the best features of a data lake and a data warehouse together.

I have read that companies have used both Databricks and Snowflake together for workloads. They may use Databricks to analyze unstructured data and then send some of this data to Snowflake.

So Databricks is good for machine learning and Snowflakes is good for business analysis.

It is much harder to build a tool for working  with unstructured data than for structured data, so Databricks tools are harder to build than Snowflake’s tools.

Snowflake though, with the emergence of AI and generative AI following the release of ChatGPT, is trying to implement tools for unstructured data, similar to what Databricks does.

Databricks is doing the opposite. They are rolling out tools to handle structured data and have created their own data warehouse.

So after about 10 years of operating in their own lane, with Databricks focusing on a data lake and unstructured data, and Snowflake focusing on a data warehouse and structured data, these companies are now starting to meet and are competing directly with each other. The competition is getting intense.

The big 3 cloud providers - Amazon, Microsoft, and Google - are also working on improving their own offerings, which they have had for a long time but were just not competitive with Snowflake and Databrick's offerings.

I still don’t see that Amazon, Microsoft, and Google’s offerings are comparable today, and I don’t expect them to be comparable in the near term, but this could change in the future because of iceberg tables.

Icebergs are the formats that both Databricks and Snowflake support.

Snowflake was rumored to be set to acquire a company called Tabular for around $600 million to help improve its iceberg offerings to compete more with Databricks, but Databricks ended up paying $2 billion for Tabular. It is rumored that this company only had $1 million in revenue, so Databricks ended up paying a whopping 2,000 times multiple of P/S!!!

Databricks must have been worried. The interest in Tabular was icebergs. Icebergs are an open-source table format that is rumored to possibly commoditized the storage of data. This could decrease the vendor lock-in effect of Snowflake’s data warehouse, and making only the analysis and querying engine relevant.

​

*****

​

August 9, 2024 - On September 27, 2022, I wrote, “WBD looks attractive here as one of the best time-arbitrage plays I see in the market today” and “at $11.32 per share and a forward P/E ratio of a little less than 6 times, the company looks too attractive for the investor who can patiently wait at least five years.”

​

This turned out to be a bad prediction and I think it really gets at the reason why mistakes can happen in the market.

The reason is simply because the future is uncertain. Buying a stock involves making a prediction on how the company will perform in the future.

My prediction was based on WBD continuing to earn decent amounts of free cash flow and maintaining, or even growing, it adjusted EBITDA, but that didn’t happen.

Free cash flow grew because of a one-off with the writer's strike where the company couldn't spend as much on content production as it wanted to. This was good for FCF during the year but hurts them beyond a year because the company now has less content to release to consumers.

What was a big surprise to me, and this is exactly what I missed, was the pace at which money earned from linear TV (networks) would contract.

Revenues for the network segments were $24.2B, $23.2B, and $21.2B for FY 2021, FY 2022, and FY 2023, respectively. Meanwhile, revenues for the networks are down 8% for both Q1 and Q2 this year, so the declines haven’t stopped yet.

And adjusted EBITDA for the networks segment was $5.1B, $3.8B, and $3.3B over the same fiscal years.

In other words, I underestimated the rate of how quickly people would cancel their cable bill for streaming services.

I just extrapolated from the recent past the high cash flows the company was earning from the networks, and while I understand it would decline, I just didn’t think it would happen this quickly.

This was a mistake. And looking back now with hindsight being 20/20, it seems pretty silly because streaming is a much superior alternative to cable.

With streaming, consumers aren't forced to buy content that they don’t want, they have access to a much larger library of content that they can watch at whatever time they choose.

Then to throw gasoline on the fire, WBD has now lost the NBA package, which will result in lower affiliate fees from cable subscribers.
 

I don’t care for EBITDA in valuation, but it is a commonly used metric when evaluating leverage so I will use it here.    

​

With all of this said, the net debt/EBITDA ratio was around 5.5x back in 2022 when the company had $50B in net debt and estimated EBITDA for 2022 of $9 billion.

​

In Sept 2022, I had estimated that the CEO had a possibility of driving EBITDA to up to $12 billion, but this was a mistake. Adjusted EBITDA for 2023 was $10.2B but could come in around $9.5 for FY 2024.

​

The good news is that net debt now stands at around $37.3 billion at the end of Q2 2024. So, using an estimate of $9.5B for adjusted EBITDA and net debt of $35 billion, assuming $2 billion of debt is paid back over the next 2 quarters, net debt/adj EBITDA would stand at 3.72x.

​

So the company has deleveraged, and it will continue to do so over the next couple years, but does it really matter?

​

The answer the market has given so far is that it doesn’t matter at all since the company’s biggest cash cow, its networks, are in structural decline with no signs of a bottom; the studio is struggling with declining YoY revenues; and DTC (streaming) is “breakeven” on an adjusted EBITDA amount, so it isn’t  really cash flow positive yet.

​

Subscribers are growing overall (domestic numbers are lumpy due to churn) and the company is guiding to adjusted EBITDA for DTC of $1 billion for FY 2025, but the market wants to see WBD prove it first.  

​

The company will be reliant on streaming for the future. I still think the company will climb out of the penalty box eventually, but I'm not buying more despite the pullback in the stock price, and I am not selling either.

​

In the end, this stock seems like it will be range bound for a while because of the large debt load, the continuing slide of the networks segment, and the loss of the NBA.

From a technical standpoint, I view it as a positive that despite the lackluster 2nd quarter, the stock price found a solid footing at $7 per share after dropping to a low of $6.64.  

​​

*****

​

November 5, 2023 - My question in my last journal entry looks like it is solved. According to an article from the WSJ, Adyen will be partnering with Plaid to help merchants accept payments directly from customers’ bank accounts.

 

If paying by ACH catches on with consumers in the US, this will help merchants avoid the high cost of credit card fees by utilizing the ACH network instead of the credit card networks (e.g.: Visa, Mastercard, Amex, Discovery), as the article from a16oz had discussed.

 

The merchants need an acquirer and a processor, and Adyen, along with many other companies, serve this role. Therefore, whether customers pay by credit card or ACH shouldn't have much of an effect on the processors and merchant acquirers. It will affect the credit card networks and possibly even the issuers since the issuers earn high fees from credit card transactions. 

​

*****

 

April 27, 2023 - I remember when Visa was trying to acquire Plaid in 2020, until the Department of Justice blocked the acquisition.

I assumed this was a good purchase for Visa because very few people understand the payments business better than the executives who manage Visa. So if they were worried about Plaid disrupting their business model in the future, then I believed that Plaid would have been a good acquisition for them, even though I wasn’t able to figure out how Plaid could possibly disrupt Visa or Mastercard.

I may have just figured out how Plaid can disrupt Visa and Mastercard though, thanks to an article from a16oz.

 
Of all unlikely places, it starts with Target.

Target created a debit card for their members called Target Red. This card is pretty successful for the company in terms of volume. It does $20 billion of volume for a company that had revenue last year of a little over $100 billion.

Here is how Plaid can disrupt the Visa and Mastercard oligopoly:  

The Target Red Card connects to the cardholder's bank account using Plaid.

Plaid is an application programing interface (API), which pretty much acts like a messenger that helps your bank and an application or website talk to each other.

Once the Target Red Card is connected to a cardholder's bank account, the Red Card can pull money from the cardholder's bank account via ACH.

ACH stands for automated clearing house, and is a network that credit unions and banks use for electronic fund transfers.

This network allows Target to avoid paying the high fees that would have gone to the credit card companies. It allows the merchant (Target in this case) to pay lower fees by using the ACH network.

 

The ACH network is owned by a nonprofit organization and deals directly with the banking network, which is the reason for the low fees compared to paying via Visa and Mastercard.

I am thinking about how this would affect another company I have been researching called Adyen.

The way I am looking at this is even though a customer is using a Target Red Card, there is a payment being made and this payment must be processed, whether it is made using ACH or the credit card networks.

 

Adyen would be the company that processes the payment, therefore, at the moment, I don’t see there being an effect on Adyen's business, but I have more research to do on this.

​

*****

​

April 22, 2023 - The following excerpt, written by Francois Rochon in Giverny Capital's 2022 annual letter, discusses why employee stock options are an actual expense: 

​

"Many Wall Street analysts and companies started reporting an additional figure to their results: non-GAAP profits which often exclude the cost of ESOs. Some financial items are non-recurring and deserve to be excluded from recurring expenses. Certain accounting charges (such as the amortization of intangibles following an acquisition) can also be excluded if it is objectively considered that they have no real impact on the underlying earning power of a company.

​

However, the cost of ESOs is real and sometimes even higher than the expenses calculated in GAAP earnings. There is a pragmatic way to approximate this cost: many companies buy back their shares to compensate for the level of dilution created by ESOs. We can thus estimate the difference between the proceeds of the issues of the ESOs and the cost allocated to the buybacks of the same shares.

​

Let’s go back to the example of Cisco Systems. For the past 21 years, Cisco has aggressively bought back its stock. It made total profits of about $158 billion over those two decades and spent $152 billion on stock buybacks. Out of 7,324 million shares at the beginning of 2002, it repurchased about 5,923 million. On the other hand, it also issued through ESOs more than 2,652 million shares for a net effect of reducing the number of shares by 3,271 million. The buybacks were completed at an average cost of $25.6, which compares favorably with Cisco’s current share price of $50. However, equity issues were carried out at an average price of $12.6. All in all, we calculated that the net cost of buying back all issued shares over 20 years was over $34 billion ($25.6 minus $12.6 multiplied by 2,652 million shares issued). This amount compares to $26 billion in total annual charges that the company expensed over 20 years. Considering that this last amount is before tax deductions, the difference between the net cost of the options and the amount necessary to cancel them was $3.6 billion. Here is a table illustrating this calculation:

​

Cisco Systems                                                                Number (m)       Amount (m)     Average Cost
Number of shares at beginning of 2022                              7,324
Share repurchases 2002-2022                                           -5,923               $151,881               $25.60
Issuance of stock 2002-2022                                               2,652                 $33,343               $12.60
Net number of shares for 2022 approx.                               4,053
Cost of option repurchase                                                                             -$34,661
Cumulative amount expensed for stock options                                           -$25,861
Income tax deductions (@ 20%)                                                                      $5,172
Difference between repurchases and reported expenses                               -$3,628

​

The conclusion is that the ESO charges were clearly real expenses. The company had to devote about $34.7 billion of its cash (or $29.5 billion net of taxes recovered) to cancel their effects, which are sums that could have been paid out as dividends.

​

At Giverny Capital, we always include ESOs in our own profit calculations. ESOs are here to stay without a doubt, but we believe firmly that we have to calculate their real costs as is done for rents, salaries or R&D expenditures."

​

*****

​

January 15, 2023 - My investment in Twilio was a mistake.

I misjudged the moat of the business and the competition.

Had I done more research through reading expert calls instead of relying mostly on one investment research publication and blogs that were very long tech than I could have realized sooner that there was other competition in this space than just Twilio.

Had I instead judged the business on ROIC, ROC and net profit margins instead of being ok with the large losses that the company was incurring, in belief that the moat was being formed and the operating losses were ok since they will be profitable in the future, then I could have realized that the moat wasn't as wide as I thought.

I thought the company had a competitive advantage due to economies of scale yet as the revenues increased, so did the operating losses.

Stock based comp was a red flag I knew about but ignored.

Had I been more patient and let the stock decline play out more then I could have saved myself from investing too soon. I recently heard a really good line from Howard Marks at an investment conference about how an investor's job is to CAREFULLY catch falling knives. I believe the safest way to do this to protect an investor from permanent losses is to stay in areas that one understands very well. I didn't understand Twilio as well as I thought I did.

I shouldn't have used the large runup in the stock price from 2020 to 2021 as justification that this was a good company with a wide moat.

​

*****

​

November 11, 2022 - On May 11, 2022, Michael Burry said the following on Twitter:

​

“Top to bottom, MSFT traded 5.2x its shares outstanding by 2002, 3.3x by 2009, and .5x so far. AMZN traded 5.7x by 2002, 6.6x by 2009, and .9x so far. JPM traded 3.0x by 2002, 5.9x by 2009, and about .7x so far. Etc. Enough takes time.”

At the time Michael said this, it didn’t make sense to me, but it makes perfect sense now.

Why would trading volume have to be so high before the stock market bottoms?

The answer is related to humans, their emotions, and volatility.

When the market is going up and volatility is low, like during the economic expansion between 2011 to 2020, it is easy for humans to hold stocks for the "long term". Rising prices and little or no economic calamities lead to low volatility because it's much easier psychologically for humans to hold.

It's when there are cracks in the economic system, like today's inflation that is leading the Fed to quickly tighten monetary policy, that make it harder to hold for the long-term.

These economic calamities cause volatility and volatility creates anxiety in us humans, therefore, causing us to trade more.

We go from "buy at any price and hold" as the market rises to "sell at any price" to just get me out and stop the losses.

And every time somebody sells there must be a buyer. So as more humans sell their stocks to lock in gains or prevent losses, each sell has a buyer on the other side.

Eventually stock prices fall from optimism or "nothing can go wrong for this company" to prices that reflect ultimate pessimism or "nothing can go right for this company".

Carvana is a perfect example.

Here is a headline for Carvana from a couple of months ago: Carvana Raised to Overweight at Morgan Stanley; PT $420

And here is a headline for Carvana from November 4, 2022: Carvana Hits 5-Year Low After Morgan Stanley Pulls Rating. [Morgan Stanley’s analyst] pulled his $68 price target and said his new base case is that the company may be worth $1 to $40 a share.

As prices crash, stocks fall to levels that most investors know are great bargains. Rather than buy though, investors wait on the sidelines because no one wants to see red in their portfolio and they don’t want to tolerate further mark downs.

This is essentially what I saw from my viewpoint yesterday on 11/10/22.

Investors knew stocks were cheap but they were too fearful to buy until there was a bullish sign. Yesterday, investors got their bullish sign - a cooling CPI number.

 

Although the CPI was 7.7%, a number that is high compared to historical standards, it is below the 9% reading that came in June.

​

Think about that, imagine someone said last year that stocks are going to be up 5% because the CPI was at 7.7%!

Investors, knowing that there are already a lot of bargains out there as it is, saw this lower CPI reading as a bullish sign that it's time to get back in and scoop up these underpriced stocks because

​

1. The CPI is cooling which means the high inflation we’ve been experiencing could be coming closer to an end and

​

2. The Fed may (key word: may) stop hiking the Fed Funds rate, and they may stop letting their balance sheet run off (QT).

All of the selling that's been occurring for this year causes lots of trading volume, and all of the buying yesterday causes more trading volume.

But what if the next month's CPI reading comes in above expectations? I'm not saying it will, but it's possible. And if it does then this is likely to lead to a lot more selling which will further increase trading volume.

So now Michael Burry's tweet makes perfect sense to me. It's the economic headwinds that leads to rising volatility that causes panic among investors, leading them to buy and sell more than usual, therefore, increasing the trading volume/shares outstanding ratio. And Michael Burry obviously sees our current economic situation as more similar to the 2000-2002 and 2008 periods than the shorter period of volatility that occurred in Q4 2018. 

​

*****

​

October 20, 2022 - Jeffrey Gundlach tweeted yesterday that when the long end of the yield curve goes flat, this could be a sign that yield increases are peaking due to "increase exhaustion". He said yields could be peaking between now and year-end because of this. I thought this was interesting. Plus I don't agree; I think yields are going beyond 5% which makes it even more interesting to me that I'm curious to see if he is correct. 

 

According to this same tweet, the yield curve on 10/19/22 was:

​

2 Year: 4.52%

5 Year: 4.37%

10 Year: 4.13%

30 Year: 4.13%

​

Still inverted on the 2-10 year end of the curve...

*****

​

October 7, 2022 - Michael Burry tweeted the following statement the other day:

"Companies that are heavily leveraged but have the cash flow and termed our debt have options today, including reducing their debt loads at a significant discount brought on by higher rates. But as [Ben] Graham said, in such a case, better off buying the stock."

I've been thinking about this quote a lot lately. 

I kept wondering why it would be better for an over-indebted company to buy back its stock instead of deleveraging in today's economy.

And I figured it out...

The reason is because the stock will only be trading at depressed levels for a limited period of time, but paying back debt doesn't have a limited time window. 

 

This works provided the company has the capacity to do it (i.e. debt is spread out with no large upcoming maturities due).

In other words, when the economy recovers and the mood swings from depressive "sell, sell, sell" at any price to optimism "buy, buy, buy" at any price, the stock price will have moved up so much that the value for shareholders won't be anywhere close to as much as before.

And if a company's stock doesn't partake in the recovery rally along with the broader market then this could def be a red flag that something is wrong. After all, the market is irrational at times but not stupid. 

Of course buying back stock instead of deleveraging depends on the company having the opportunity to do this.

For example, I wouldn't do this if I was managing Warner Brothers Discovery because they have a lot of debt due in the next three years. Therefore, it makes much sense to pay back the debt instead. In other words, it's better to pay back the debt and make sure the company lives to fight another day than try to be a hero for shareholders.

If I was management of AB InBev then I would be buying back stock right now. The company has prioritized organic growth, deleveraging, strategic M&A over capital allocation, but at these current prices, it makes more sense to buy back the stock.

 

I would even scrap the dividend to buyback the shares instead. The company can afford to do this because their debt is much more spread out. They only have $3.2 billion due in the next 3 years and around $12.2 billion due in the next 5 years. The debt due in the next 5 years seems high but free cash flow generation is strong enough to pay this off.  

So in summary, I see a lot of value in Benjamin Graham's statement of prioritizing buying back the stock with free cash flow instead of deleveraging if a company has the ability to do it because there is a limited window of time when the stock will be a available at good prices, meanwhile, the debt can be paid back anytime before maturity.

​

*****

​

September 27, 2022 - WBD looks attractive here as one of the best time-arbitrage plays I see in the market today.

The company is hated by the market because the company is in turnaround mode after being poorly run by previous management at AT&T.

The current CEO, David Zaslav, is shelving projects (Batgirl), firing people, transitioning to direct-to-consumer by combining both HBO Max and Discovery Plus, all while continuing to run a slowly dying linear TV business and a movie studio.

Not to mention the company has a lot of debt, $52.3 billion gross as of 6/30/22, and net debt of $50 billion.

There is some hope though but it will take time for the company to be turned around.

David Zaslav previously merged together Scripps Network with Discovery so he has some experience here. And with Discovery and Warner Brothers, he has some levers to pull.

The company has probably the best (maybe 2nd after Disney?) intellectual property in the media industry.

Adjusted EBITDA came in below expectations last quarter but there are lots of synergies that could bring it from a little more than $9 billion (2022 guidance) to $12 billion next year.

For one, CNN Plus probably cost the company $100s of millions of dollars, but that was cancelled so that will save them money next year.

And of course eliminating duplicate positions and overhead will be gone as well.

At $50 billion in net debt and at 2022’s guidance of around $9 billion in adjusted EBITDA, the expected net debt to EBITDA ratio for 2022 is running at a very high level of 5.5x.

But if the CEO can get adjusted EBITDA to $12 billion, which doesn’t look like that difficult of a task, then the leverage ratio falls down to 4.1x without paying any debt.

But they're expected to pay down debt so if the company can pay down $3 billion in debt this year and $3 billion next year then the net debt to adjusted EBITDA ratio could drop to:

2022 net debt/EBITDA: 3.9x
2023 net debt/EBITDA: 3.6x

​

This leverage ratio is much more reasonable. The company has $6.5 billion due in the next three years and although I believe they will pay it off, or at least refinance, this is what would scare me the most since this is a time-arbitrage play with time and patience being the most important factors here for investors.

​

$6.5 billion in the next three years is more than what I would want to see due in such a short period of time, especially in such a difficult macro environment.

But at $11.32 per share and a forward P/E ratio of a little less than 6 times, the company looks too attractive for the investor who can patiently wait at least five years.