Cloudflare’s use of Databases

Tim Green at TMF wrote a good article about Cloudflare’s announcement yesterday regarding Cloudflare now allowing outbound TCP connections…?

Up until now, Cloudflare’s Workers did not support TCP connections. That means that a developer running a database somewhere else could not use Workers with that database without resorting to a middleman. A developer could set up a server application outside of Cloudflare that accepts API calls from a Worker, pulls data from the database, and returns it to the Worker, but the Worker could not access the database directly.
Database access was the missing piece of the puzzle for Cloudflare workers. Once a Worker can connect directly to any database, the potential of the platform greatly expands. Suddenly, a developer hosting a database on AWS, or on any other cloud platform, can run their actual application on Cloudflare Workers, benefiting from the global reach of Cloudflare’s network.

I’m still waiting for Cloudflare to unveil D2 to replace their underwhelming D1, database. Then we’ll have a whole R2D2 ready to go, right.

Better late than never,



Seems like such a basic requirement… Wonder why it took this long. Also, wonder what other similar basic requirements are unmet and floating around?


On another thread, Saul asked, “ But I’m still not clear why you are overweight in Cloudflare, except “hoping” for a turn around, I guess.”.

I’m putting my response here because I see Cloudflare’s puzzle pieces (combining S-curves) coming together with this and some other recent news. In this context, is it unreasonable to believe Prince has waited to provide such a rudimentary feature, in keeping with his grand vision, when we know he’s been building out the primatives for Workers for years and has only recently been ready to go GA.

I’m looking at the long range vision of the Founders. I mean aren’t we investing in Primarily founder led companies for this very reason. Through thick and thin a Founder is more likely to do what’s best for the company in the long run, right.

I understand that hope is not an investing strategy we follow here. And we don’t invest in turn arounds. But, when revenue growth goes down and really it’s more to do with macroEconomic factors than execution, I don’t believe it’s a turn around.

Cloudflare showed better growth endurance than

  • Monday: turned in 50%% after growing at 68% → 50/68 = 73% endurance
  • Cloudflare: turned in 37% after growing at 49% → 37/49 = 75%)
    I’m not saying that Monday didn’t turn in other numbers that were much better. They did. I’m just saying that Cloudflare isn’t, IMO, a turn around story.

I see each company consisting of multiple puzzle pieces. When MacroEconomics is as strong as they are now, in influencing the revenue growth, I’m focusing more on other aspects of the business. Sure, the business model of SaaS provides the benefit of ramping up the bottom line, even during times like these. But, I am counting on the Founders to keep investing in the company through thick-n-thin.

I made some assumptions when investing in many of the companies discussed here over the last 3-4 years. I was drawn here mostly because I understand what Saul kept saying, “It’s impossible to time the market.”. What this means to me is that I bought mostly founder led companies because I saw the long term vision of those companies becoming behemoths.

My investment thesis in Cloudflare is stronger now than ever.




This is a question that has been weighing on me for some time now, and I can’t find an answer in the Knowledge Base. What, exactly, defines a “turn-around” stock? This may be OT, but since it’s part of the rules of the board not to include “turn-around” stocks, I feel like I need more definition.

I would assume, for example, that a company that has filed for bankruptcy, restructures, and comes back to life is a turn-around stock. But suppose new leadership really has–well–turned them around? I don’t have a specific example to toss out here, but it’s conceivable that something goes back to hyper-growth, or maybe hits hyper-growth for the first time after a bankruptcy. Do we not talk about that stock if it’s hitting 40% growth now?

What if a company does not go bankrupt, but goes sideways for a few years. Or goes down substantially in revenue–it exits hyper-growth for a time. How many quarters does it get before it becomes OT as a “turn-around” stock? What if they’ve been failing miserably and a new CEO comes in and turns things around? (I’m watching Fastly in this regard.)

Are the only companies we can consider here ones that never faltered in their history? Only ones who never have a series of bad quarters even during bear markets, recessions, or black swan events?

Every time I think about that question, Cloudflare and Prince’s prediction of “stacking S curves” comes to mind. For the sake of argument, let’s say that’s an accurate prediction. To have stacking S curves, you have to have the downward slopes of the S as well. After two or three S-curves, you could pull back the lens and see a larger pattern of maybe a bigger S curve–higher highs and higher lows in each successive S, just like trading charts have patterns within patterns.

But at the outset, you just have a company coming down off their first S curve and you have to judge whether the pipeline and management team can deliver on their promise to start another S-curve. How long do they have to do that before the next S is considered a turn-around?

I’m not trying to be difficult; I truly do not understand how “turn-around stock” is defined here. And, even if there is some kind of definition, I don’t know why a hypothetical stock that fell on hard times for awhile but then came roaring back into hyper-growth would be OT.




Thanks for the question.

Wanting to buy a stock that has cratered in a normal market because the underlying company has had bad or even disastrous results, and you want to buy it because it is cheap and you “hope” it will go back up, is a “turn around” hope and is OT for our board.

Buying a stock that is going down in a difficult macro situation, but for which the underlying company still has exemplary results, even if the results are less than they were six months ago, is not OT for our board.

Buying a company which is “roaring back into hyper-growth” is, of course, not OT for our board, and I really can’t imagine how you could have thought that.

In other words, when talking about investing in a turn-around based on a hope being OT, we are talking about the company’s business, not the stock price.

I hope that this clarifies it for you.


PS - And I see that Will02028 already defined it clearly for you in the very quote that you used:


Thank you, Saul!

I didn’t see the section I quoted as an answer to my question. Jason began the post I quoted from with:

On another thread, Saul asked, “ But I’m still not clear why you are overweight in Cloudflare, except “hoping” for a turn around, I guess.”

So I read the section I quoted as Jason’s response that he didn’t think Cloudflare was either of those things. Since it appeared that the two of you were debating that very issue about Cloudflare, I wanted to back up to get a definition.

Either way, though, you’ve answered my question, for which I’m grateful. Since I only started investing in 2021, I’ve never encountered a “normal” market, so it’s hard for me to sort these things out!



I think Jason successfully defended Cloudflare as “not a turnaround.”

The word I’ve used is “speculative.” It seems like a lot of investors believe in what Cloudflare will do – in the long term. E.g. the “stacking S curves” thesis, which hinges on how significant (measurable in future revenue) nascent and future “Acts” will become.

There’s nothing wrong with speculation about the future. But the more grand and long term your vision, the less you’ll be able to “follow the numbers” like many of us do here. How will you know if your thesis is playing out?



Of course. The article frames it as “fixing” and “proper database support” but the scope is way beyond that.

Currently, there is no standard API for socket connections in JavaScript. We want to change that.


When we decided that we would add the ability to open and use socket connections from within Workers, we also agreed that we really have no interest in developing yet another non-standard, platform-specific API that is unlike the APIs provided by other platforms. Therefore, we are extending an invitation to all JavaScript runtime platforms that need socket capabilities to collaborate on a new (and eventually standardized) API that just works no matter which runtime you choose to develop on. […]

Once released, we expect Socket Workers to blow the doors wide open on the types of intelligent distributed applications that can be deployed to the Cloudflare network edge

(, November 2021)

They then – in partnership with Vercel, Shopify, and others – also went on and established a new working group focused on interoperable implementations of standardized web APIs in non-web browser, JavaScript-based development environments. Because, well… that didn’t exist either.

Flash forward to today, you can now use seemingly rudimentary a thing called connect(), which may or may not become de facto standard in the future. The initial focus has been on enabling more efficient ways to connect to databases on the backend.

It’s not like a situation where a certain other company still seems to be struggling to meet the basic requirement of full IPv6 support… ( :wink:

Regarding “But I’m still not clear why you are overweight in Cloudflare, except “hoping” for a turn around, I guess.”.

As the recipient of said inquiry, I found it to be neatly worded, and appreciated it. After all, it should be far from clear. The earnings release was quite a shocker and my first thought was “there better be a really good explanation for this”. Then came a jumbled mess of an earnings call. Alas, I didn’t have time, nor energy to lay out my take on the quarter.

Part of my reasoning is implicitly laid out here:

And - after spending a weekend deciphering an earnings call - here:

Then investor day came along, with a tremendous amount of information and color.

My conclusion was that the challenges faced weren’t specific to NET, and I ended up with lowered near-term expectations but simultaneously with increased confidence. I find them to be in a strong position - increasingly so.

There’s a slowdown across the board. Jamin Ball paints a good picture with several charts, showing growth, NRR, net new ARR, and more.

My take on some aspects of the quarter, in short:

  • Rapid and sudden shift in macro caused significant elongation in sales cycle, and with NET’s rapid sales cycles, effect and uncertainty shows up quickly. This is the reason for a downward revision in guidance.

    • More specifically, larger deals involving larger projects seemed to come to a screeching halt. Described in their Q1 call as “large project around moving to a Zero Trust environment or something else that would be a larger initiative.”
  • Go-to-market talent changes isn’t anything new, i.e. it was already factored in when giving guidance back in February. From their Q4 call: “In our guidance, we have not factored in any improvement in the macroeconomic environment or from our go-to-market initiatives.”

  • Improving their GTM is a positive (i.e. room for upside) and a project that began more than a year ago.

    • From investor day (rough transcript):

      (CEO) “…sales team, we owe a ton to Chris. Very few sales leaders go from zero to a billion dollars in revenue. But if we’re again, super honest, Chris, he got to the Peter Principle and was not making those changes. So we needed to change the person in order to change the team. And we spent a bunch of time, again, finding the right person […] Mark to do this. […] Very first thing that both Michelle and I said to Mark was there going to need to be substantial changes on the team. It’s going to take you a bit of time to of get your hands around it, but I want your kind of 120 day plan to be like, how are you going to execute it? And almost on his 120th day, we pulled the trigger on executing that.”
  • Presenting these things together wasn’t the brightest idea, so they used part of Investor Day to
    bring clarity on this and a whole lot of other things.

    • From investor day about why they brought up GTM on the earning call:

      (CFO) “The transparency we share with you is also something we share internally. And we wanted to be in a position to talk to employees fast and not have rumors and information and misinterpretation out there. And we thought by making it public on a platform of an earnings call would give us the ability to do what needs to be done with speed and transparency in front of the employee. That’s why we put things together. Little did we know that it would be combined. If we had spent more time thinking about it, we might have come to that result. So that is clearly on us.

      “But it has two reasons we wanted to be fast on the execution. North America will happen a lot faster. So that might be executed by the end of next week already. Europe might take a little bit longer. And I think if you look at the model and how we build it, ramping up new AEs is very efficient. Within Cloudflare, we normally get a mid-market AE to full productivity in four months, an Enterprise AE in six months. That was another reason why we wanted to be fast, because if we do a good job managing through this upgrading talent, we might see impact, positive impact, before even the year is over.”

I encourage anyone interested in NET to watch the Investor Day webcast(s), especially the investor session ( It touches on a lot of what’s been discussed here since earnings. It’s a long webcast, so to save some time:

  • ~1:04:00: Marc Boroditsky - Go-to-Market
  • ~1:26:00: Thomas Seifert - Macro, guidance, metrics, unit economics, and more
  • ~1:48:00: Q&A session

I see NET as the canary in the coalmine, or a coalmine. Companies with longer sales cycles don’t see impact on revenue as quickly. So the question for me boils down to “what will macro do?”. Is this a temporary blip? Will it get worse? To which I have no answer.

Cloudflare is still my highest confidence holding, and that’s based on my assessment of how the company is doing, and not on whatever the stock is up to.


Compliments are rarely given in a reply post, ‘unless it’s really deserved’, paraphrased from Saul.

Raylight, I just really appreciate all the work I know it took to summarize the last 1+ years of innovation, regarding the update starting this thread.

Thank you.

As far as my reply to Bear, about calling Cloudflare ‘speculative’. Wow, I knew you were focused on creating statistics around the numbers given at Earnings; but, I didn’t know that’s how you’ve viewed the rest of the process a lot of people here go through, when they’re evaluating a company’s progress.

IMO, all we are ever able to do, as investors, it best guess the future.

In addition to my A grades in graduate level Statistics, I was an ‘Approved Tudor’ for Calculus at my University, which is primarily the way in which approximations are quantified. I don’t say this to impress anyone. I’m say this to perhaps give substance to my reasoning, that both of these ways to quantify potential outcomes, once given, are then very much up for interpretation.

There was a time, on this board when dozens of posts were dedicated toward quantifying Saul’s wisdom. All to no avail, as I recall. I participated in this and was proponent of Saul’s ‘modified’ Oomph equation, as long as the very long list of other factors are no ignored.

I don’t believe your post is asking for a renewed visit into how best to quantify our way out of ‘speculation’. I am curious, where are you in your view of the last three years. Was your focus on statistical models of the numbers an attempt to avoid being speculative? What weights are you giving to the other areas in Saul’s knowledge base?

I posted my attempt to give weights to the various aspect of Saul’s knowledge base. I won’t post it here; but it grew out of one of my monthly summaries two or three years ago. It was 1-7 point scale, quite an arbitrary number. I never followed up, to refine it. I’ll be honest, I believe the sub-conscious mind does a fairly stand up job mashing together all of the inputs, once properly absorbed (properly absorbed refers to the Slow part in the Thinking Fast and Slow; I submit no one should underestimate the value of the Fast part, after doing this work).

I do get frustrated trying to explain my process due to the last paragraph. I include to be intellectually honest.




Hi Jason

Regarding your comment that you feel the unconscious mind does a decent job of reaching conclusions, in case you haven’t yet read it I ‘highlt’ recommend Thinking, Fast and Slow by Daniel Kahneman. You might revise your position…


I still have 15% of my portfolio in NET and have only trimmed about 5% of my total shares since January 2023. Did not sell anything at all after earnings call.

I believe Bear’s thought on “speculative” is that Cloudflare is slowing down and we are betting on it turning parts of its business around - be it that the sales org rebuild will lead to another round of growth, or one of its new products will become a major revenue driver. NET once had a durable revenue growth, and now it doesn’t. He doesn’t want to bet on that and I respect that.

I believe that the board members all have different investment horizons and we need to take that into consideration when reading their thoughts.

For example, I have a longer time horizon since I’m in my 30s and I have ample amount of income outside of investments. I believe that NET’s products will lead to success. The amount of information they have about internet will also lead to products we have not thought of. I have no idea about timeline. I also believe that the stunted growth is a short term effect. I think their sales org needed a reboot and I’m willing to wait it out. Why? Maybe because I have a different investment horizon. If they continue to grow at 50%+ quarter over quarter, I may put 25% of my money in it. But since I have no idea what their growth is going to look like, I kept it at 15%.

I was disappointed at their Zero Trust release about AI the other day. I think this was a press release and not much of a value add to companies that want to get better at AI. But who knows what else they are developing. One of the products I already saw adoption is the Tunnel product (reverse proxy). They already have R2 which will be invaluable in the long run. I’m willing to put 15% of my money on it. But it may be speculative if you are looking at their growth numbers in the next few quarters. I am ok with that.


Sorry all – I don’t know how I could have missed the fact that WillO actually mentioned Kahneman in his post!

Although it is a bit tangential, one of Kahneman’s points, in my opinion, is that ‘slow thinking’ is the antithesis of relying on subconscious thinking.

1 Like

Makes sense. And maybe if growth continues to slow (to say, less than 20% YoY) you’ll reduce from 15% to 5%. I actually don’t think that is speculation. You’re watching how they grow and reacting to it.

My question for others is, how else do you evaluate how the business is doing? It’s great to announce cool products…but when do you decide whether or not they’re working (and if they’re selling)? Is there any way to track the progress of your “long term” investment?



This is a great question and one that I struggle a lot with. We can look at product announcements, new customer adds, ARR, etc. but it’s really tough to build a “model” of how that will translate into growth in year t, t+1, t+2, etc. How many of these factors are true signal (and with what lag), how many are collinear and providing the same information, and how many are just plain noise? On top of that, how much of this information is already baked into the stock price because others are looking at the same information?

This may be slightly off-topic, but personally I’ve started relying more on valuation using analyst forecasts (which are obviously imperfect, but they’re building the “model” for you with publicly available info). If you can build a simple DCF with their assumptions and generate a healthy return then that means there’s room for error if the macro slows down or the company stumbles with it’s GTM strategy. If a company is already priced to perfection with all the publicly available information baked in, then the stock will be punished for any misstep and you are relying more on growth that the majority don’t see (or maybe they “see” it with you, but it can’t be quantified yet).

This obviously doesn’t mean I rely solely on this–I’m picking the stocks to even evaluate–but within a basket of potential stocks it’s helpful for sizing, choosing when to buy / trim, and managing expectations in the shorter-term. Two good examples were CRWD and ZS over the past couple of months. When ZS was in the $80’s recently, I wasn’t feeling great about holding it, but my simple analysis was showing really high price upside over the next couple of years just using analyst growth expectations.

While sentiment and momentum were poor because of slower billings numbers (or competition, etc.), that information was likely already being priced in (plus some) and just a reversal to the mean would generate nice returns from that point. It also wasn’t like ZS was a broken company or value-trap–maybe it would grow 25-30% instead of 35% over the next 2 years, but it was being priced as mid-teens grower. After their earnings pre-release, the stock has been on a tear because expectations were so low.

NET is a tough one since it’s always performed poorly on these valuation exercises due to the expectations of future innovations / revenue, so you have to believe in the future growth story in order to own it (which I still do, but less than before). Ultimately the growth of most of the stocks discussed here will provide returns over a long enough timeline, but the further out the potential returns are, then there will be higher execution expectations / associated risks, which requires conviction in the management, TAM, “story”, etc.


I ponder over this a lot. The part we are all trying to figure out is longer term growth durability. The current growth numbers tell us that something is working for the company right now. They don’t tell us if trend will continue for next year, but the current growth numbers gives us some conviction. DDOG, SNOW, NET had great numbers last year, but their revenue has decelerated significantly this year. How do we know the same won’t happen to the current high growers and DDOG, SNOW, NET won’t stabilize next year. The older times with FSLY, etc in a bull market was different as it was clear the story had ended.

My point is we are all speculating to some extent on next year growth based on bunch of data. It’s essentially a bet on future growth, but I do agree on position sizing to minimize risk.

I am long SNOW, NET and DDOG.


My point here has been that Cloudflare is different, because you’re not just speculating on how the business will grow, but on how new businesses (lines of business) they’re entering (or have entered, or will enter) will ramp up, and if they’ll be able to sell these new products, and how much.

It’s not like Datadog or Crowdstrike when they come out with a new “module.” They have a long history of doing that, of succeeding massively, and it contributing meaningfully to revenue growth.

Cloudflare has a history of the new products, cool as they sound, not contributing meaningfully to growth, and apparently they’re also having problems selling them as they told us in Q1.

I’m not saying Cloudflare isn’t worth speculating on. I’m saying speculating on new lines of business is different than speculating on growth in a current line of business.



There is another issue with the whole “multiple new products every quarter”. I work at a large tech company and we’ve launched new products and set salespeople on a wild goose chase, until we realized they had no idea how to sell them. They were very different in nature from the stuff they were used to sell and they were confused. They had training, but they didn’t quite understand how to position the products to customer business challenges, or even the technical problems they would solve. People forget that most tech salespeople are not as technical as we would like them to be. We had to change the strategy.
Once we built a whole new product and realized the buyer wasn’t IT, but HR. And we never sold anything to HR before. It took a while to create those relationships.