Tech Bankers ... How Technical Are You?
I am currently gunning for tech groups only as I am a CS + Finance major but have just begun to realize how much of the technical shit that big tech giants (MSFT, AMZN, GOOG) and SaaS startups (SNOW, MDB, DDOG) actually do at a deep technical level. I feel like understanding the nitty-gritty technicals determines how accurate your assumptions in 3-statements and rev builds and can be, as well as how comfortable you feel with working with the details of a super technical company when preparing marketing materials and CIMs in an auction, and was just wondering for those in industry how deep your technical understanding of EE/CS (mainly CS though) really is.
I did 3y of tech banking at a EB (global top bucket every year) and have 0 technical knowledge. You’re a banker not a software engineer, you need to understand the business drivers, what investors like (modern cloud product, recurring revenue, low tech debt etc) and tell the right story. If you look at the fee opportunity majority of it is in software - software PE investors are just some more people like me who have studied finance and left banking.
The “technical” section of CIMs etc is mainly lifted from company materials, there isn’t a deep amount of value add.
There are some recurring concepts that you learn (eg what I mentioned above) but it’s nothing you can’t learn by spending some time on chat gpt.
There’s some areas of tech which are more technical (eg things that are a bit more hardware heavy or less cookie cutter software) where having prior studies helps but it’s a very small niche
I don't know how to code or actually utilize any of the software or infrastructure hardware (servers, storage, networking) that my tech clients provide.
However, I do know the tech concepts at a high level, so I can understand what drives valuation and financial performance, and when developments occur in the industry (like M&A), I can understand why it happened and how it impacts companies in the space.
For example, I know nothing about how Red Hat actually works, but I can explain IBM's acquisition thesis in detail and how Red Hat fit into IBM's strategy to transform its business from a legacy mainframe hardware company to a hybrid cloud infrastructure company with significant software mix. I understand the drivers of the strategy at a high level, i.e., hybrid cloud was becoming increasingly important as public cloud providers like AWS, Azure, etc. began seeing massive adoption, but most companies still had on-prem infrastructure, so hybrid cloud management was becoming more critical.
I understand the general adjacent sectors and the value chain, so I can recommend acquisition targets and understand the thesis behind deals that are done. For example, I don't know how ServiceNow or MoveWorks actually work, but I understand that ServiceNow traditionally handled IT support tickets, and those are traditionally a pain-in-the-ass for everyone, so MoveWorks helps ServiceNow develop an AI-driven function to streamline the ticketing function. Now if I'm advising a ServiceNow competitor, I can speak to this deal and recommend targets in a similar space.
I understand the business models work, so I can see the logic behind certain SaaS companies like Datadog offering their services through a consumption-based pricing model as opposed to a traditional SaaS subscription, i.e., because they host the software on their own infrastructure on behalf of their clients, so this helps account for clients that are costing more because of higher data usage. That allows me to understand why these companies de-emphasize ARR in favor of metrics like RPO and revenue (because revenue recognition isn't actually guaranteed and depends on when the customer actually utilizes the software). That's helpful when I'm advising clients on quarterly earnings and what metrics to disclose.
Just a few examples, but the point is you need to understand the tech at a basic level to really have a solid grasp of what drives your client's business, financial, and valuation performance. You don't need to learn coding or learn how to use the tools your clients sell, but you need to understand the use cases, business model, and "value chain" in terms of which companies/subsectors they depend on and which companies/subsectors depends on them.
With that said, we are a buyside-advisory boutique adivsing large tech clients on retainers, so we think a lot more broadly about strategy and are involved in tasks like M&A targeting, quarterly earnings preparation/setting guidance, thinking about how to drive value creation (increase share price) organically or inorganically, etc.
Perhaps you don't need to know as much if you're focused on sellside advisory since you're in more of a process-management role and you redirect most buyer questions to your client management team anyways. Certainly felt that way when I was an intern at a sellside-focused tech boutique. Not sure what it's like at BB/EB, so I defer to others on that.
>“For example, I know nothing about how Red Hat actually works, but I can throw six paragraphs of consulting platitudes and buzzwords at you until you beg me to stop talking”
Fair enough. Let's cut the platitudes and buzzwords.
Companies used to have their physical server and storage boxes in their own physical spaces, either in their office or a private "data center", completely owned by the companies. These server and storage machines do all the computing stuff that keeps a business running, but they are big and expensive to replace and maintain. Upgrading was also disruptive because you need to rip those boxes out and replace them physically.
However, public cloud companies like AWS, Azure, etc. began offering companies the ability to rent these machines and access them over the internet. Now companies don't have to rip down their old boxes in their space to install new boxes. They can just rent more or less through the public cloud and upgrades are handled by the big tech companies like AWS.
But these are very large boxes and machines! There's also a lot of important stuff running on them and it's hard to move them all to the public cloud without disrupting your business.
Therefore, companies ended up having boxes in their own physical space as well as shared space hosted by the public cloud. This was called "hybrid cloud".
You could have software developers working with the boxes in the physical space and other software developers working with the rented boxes in AWS's physical space. If the two teams are working on developing the same software application, it needs to work for both of them across both environments as they collaborate.
Red Hat helps the employees and boxes in any location work together smoothly. Rather than hiring a huge, sophisticated IT team to oversee where / how data and applications get stored and accessed, Red Hat automates many of these functions so people, especially software developers, can work together and access company data and software applications running in different environments.
Therefore, as more companies ended up in this situation (hybrid cloud adoption), Red Hat became more important.
Red Hat was also big, all software (no hardware), and gave IBM a faster growing recurring revenue stream.
Therefore, since software comes with better gross margins than hardware, and recurring revenue decreases volatility (more predictable), IBM saw their P/E multiple expand over time because investors rewarded their valuation as they successfully executed on this strategy, as can be seen in their margin expansion and stable revenue growth over time.
That better?
Edit: Better wording in one of the paragraphs
Comments like this are why I still check this website. Very thoughtful and same goes for the response to the troll below.
would you recommend to someone to make the move from BB sellside advisory to buyside advisory if they are interested in a better understanding of how companies create value, and learning how to evaluate decisions from a corporate strategy standpoint?
Del
you just pull consensus numbers or use management assumptions
you're not doing any real analysis of the business beyond what you've been provided. If you were going into equity research / publics you'd have to be more familiar with the space, albeit after you're hired is fine.
I know enough to know that software as a service is a scam.
Why?
I totally get what you're saying, and you'll understand this if you majored in CS and did tech work.
I'm gonna build on top of an example mentioned here: Datadog.
I recently had the chance to talk to creators of Husky (Datadog's blog has detailed posts on "Introducing Husky" and "Husky Query Architecture") and read their blogs. When they mention trillions of events, I totally get what they're talking about. At work we use Splunk. Our application has ~1 million visitors per month, with each session having 5-50 logs/events recorded to completion (so at least 5 million events for our app per month). Each log has data about the date, user, device type, error or log, app ID, etc. This ends up being a couple of gigabytes of data per month. When we need to debug, we have to query all of these logs and sessions and figure out what is going on, and a simple query sometimes takes up to 5 minutes only over the last week's data.
Building an infrastructure that supports optimal storage and querying of such a large volume of logs is really complicated (you need to figure out how to shard and distribute - companies pay top dollar for distributed systems SWEs, some of the biggest software challenges is all about distributed systems and this shows up very often in streaming like Netflix and Youtube because you want to keep latency low - and what to cache and so on). Along with the dashboard and etc., it's really complicated that your enterprise or medium-size company cannot single-handedly pull it off, and it's not really worth the investment to create multiple teams to just build this - we are talking trillions of logs for an enterprise where sometimes the events are interconnected (you don't want your query for one type of log to take 30 minutes or something). I also recently attended a tech conference and got a firsthand demo of Datadog software from their tech sales team, and at work I have used Splunk, so I have a good idea of how they work.
I have also seen how embedded Splunk is in our company. We have multiple wrappers on top of it. I actually made one of the wrappers for our log system, and it's nearly impossible to switch out from the software once it's integrated, so they have a strong tech moat and high switching cost. There are also people are subject matter expert on Splunk that we get help from and their entire job is just about Splunk, that is what usually Site reliability engineer does.
There is also the culture aspect of tech companies. For example, Cisco acquired Splunk, and Cisco is not a great tech company with a good culture, while Datadog both pays better and hires better talent, so overall I'm more optimistic about the future of Datadog. You can kind of see this from the data already:
Revenue Comparison (Latest Full Fiscal Year Data)
Metric
Datadog
Splunk
Annual Revenue (FY 2024)
$2.68 billion
$4.22 billion
Year-over-Year Growth (FY 2024)
26.1%
15.4%
I'm also seeing Datadog putting more effort into integrating AI. Although I think Splunk's UI is better. But overall, they're both great companies, very sticky, hard to replicate, and companies of all sizes need observability tools. There are also upcoming tools that offer generous free plans to get users, and they are actually easy to integrate to apps. I also recently had the chance to try Betterstack myself and it's decent for a small app use case, but I don't think it's suited for a big company yet.
But what I'm trying to get at is that to be a successful salesman and investor, you need to first understand the product value prop, market fit, technical side and its moat, market size, interact with the builders and operators, and understand potential expansion points. Then you can dive into numbers like revenue, cost, profitability, customer acquisition costs, data storage costs etc., and you'll have a much better idea of what the value drivers are. Everyone knows the numbers, but the nuances make the difference.
Also you can now do a bottom up exercise to figure out the revenue and potential revenue and estimated customer count and size and costs.
You wrote all this $#it on thanksgiving after stuffing your face with turkey and watching the games?!?…I can’t even read it never mind write it…Respect!
HAHA, I got a chuckle reading this, such is life. Honestly just really passionate about science, tech and finance.
May I ask what you do for work? Sounds like you’re in the business, so curious your thoughts on the following:
My understanding was that Splunk is a more modular/flexible observability solution relative to Datadog/Dynatrace and made to handle both cloud and on-prem environments.
It’s more complex and requires a more sophisticated IT / data analytics team to operate, but in large enterprises which generally have more complex hybrid cloud environments, it’s cheaper to run at scale and they generally have the engineering talent already to keep it running. It also allows for more data sovereignty since a company can self-manage Splunk on their infrastructure. The downside is it’s difficult to operate as you mentioned and requires teams dedicated to running it.
I think for that reason, Cisco acquiring Splunk made sense. It’s less about the culture fit of the companies and more the customer base they serve. Both are targeted towards large enterprise deployments that are often hybrid cloud. Cisco acquiring Datadog for example would make no sense, no? A public cloud company like Amazon (AWS) / Microsoft (Azure) / Google (GCP) / etc. sound like the more logical acquirer (and maybe SaaS giants like Salesforce), would you agree?
Datadog on the other hand is much more “plug-and-play” and “cloud-native” meaning it’s meant to run on Datadog’s infrastructure (likely hosted on AWS/Azure/GCP I would imagine, though not sure) and fully managed by Datadog. Even though you can install Datadog agents on your on-prem infrastructure, that means you have to trust Datadog with all your data and backend processing, which is not only inefficient but also not a possibility for highly-regulated industries like banks or defense companies with compliance requirements. It’s also harder to control the cost as the data usage of an on-prem environment managed by Datadog’s cloud-native platform is unpredictable.
Datadog is more suitable for organizations such as SMB enterprise SaaS who have a limited, fully cloud-native environment. However, in practice, global companies could very well use both in different parts of their company.
Some parts in one geography may be fully migrated to the cloud while another part in a different location may still be running a hybrid or on-prem environment. Therefore, even though Splunk is better for large complex enterprises, as enterprises continue to migrate workloads to the cloud, Datadog is able to take some share-of-wallet within those enterprise customers while also taking share of end markets like enterprise SaaS startups which have obviously proliferated over the past decade. That would be the driver of their faster growth profile compared to Splunk.
Would you agree with that characterization?
SWE.
That is a good point Splunk does offer on perm while Datadog does not that is perhaps why our company is using it. GPT answer: 'Datadog does not do on prem because its whole product, go to market and unit economics are optimized around being a single multi tenant SaaS control plane. Splunk does on prem because it was born as software you install in your data center and it still has a huge legacy customer base that cannot or will not move everything to vendor hosted'.
The reason I mention culture is because it has long term impact on company, if culture and pay declines you have slow quitting which impacts your company down the line.
While technically MSFT or Amazon or Google are potentially better fit to acquire Data dog as they can utilize their servers for storage, the valuation is straight up too high, 60bn is too steep. All the big tech companies have built in house observability tools for their apps, so the I don't think the value prop for them is there to pay so much. I think they sell it too.
yeah, I agree with your analysis. Although, Netflix is an example of a big company using Data Dog and they are fully on Cloud. Also Capital One has used Datadog in the past. I am also gonna say this statement does not seem right to me:'Even though you can install Datadog agents on your on-prem infrastructure, that means you have to trust Datadog with all your data and backend processing, which is not only inefficient but also not a possibility for highly-regulated industries like banks or defense companies with compliance requirements.', even if you use Splunk you are still going through their platform and regulation is not a barrier to these types of stuff.
EDIT: Oops, meant as a reply to @regret_tech
First of all, appreciate the constructive dialogue. Hard to have these convos on WSO these days. A few more thoughts below. Welcome any other points of view if you have any.
Totally agree with the synergy thesis between Splunk and Cisco. Think we're in complete alignment.
The point I was trying to make was that I see limited synergy opportunity:
(a) Between Datadog and Cisco because of their conflicting strategies around cloud-native vs. on-prem, respectively, and
(b) between Datadog and other large cloud-native SaaS companies like Salesforce because they target different buying centers (i.e., Datadog --> technical teams like SRE, DevOps / Salesforce --> operational teams like sales & marketing) or are too small to execute a deal of this size
--
To your second point, again, totally agreed. Mature enterprises, by definition, have been around for years before "cloud-native" emerged. Very likely, significant portions of their "mission-critical" applications are hosted on-prem. Of course there are exceptions like Netflix, but the majority of large, mature enterprises are still operating in hybrid environments (I believe...not entirely sure what the data currently says).
Part of it is cost control / vendor diversification, but I think the bigger reasons are constrained IT budgets, unwillingness to undergo a massive potentially-disruptive cloud migration, and critically, compliance with data governance restrictions.
--
That brings me to your last point. To clarify, the concern isn't so much around the type of logs / type of analytics being done on the logs and more about the security / control of the infrastructure itself.
With Datadog, your workloads are running on servers in shared multi-tenant data centers managed by Datadog. Historically, that was unacceptable for some government agencies (definitely not all, but some) due to the security risk of shared environments without sufficient isolation.
That's not to say the government isn't making progress with cloud adoption, they are just doing it in a slow, deliberate way. FedRAMP exists for exactly this reason. Splunk achieved "High Impact" authorization in September 2024. Datadog still isn't quite there yet though they are in-process.
Splunk previously achieved Moderate Impact in 2019, whereas Datadog didn't reach this level (still currently at Moderate) until 2022. The VA actually got in trouble in 2019 for using Datadog before they were authorized iirc. I don't think I'm allowed to link on WSO, but it's all public and easily searchable.
As enterprises and other customers, like the government, continue to find ways to migrate to cloud, they are becoming more receptive to newer cloud-native vendors like Datadog, and therefore, Datadog will likely continue to expand their share-of-wallet among large enterprises traditionally dominated by Splunk.
It's not that they all of a sudden outcompeted Splunk for "Moderate Impact" level government teams post-2022. They just didn't have access to this market before, whereas Splunk was already there.
Additionally, Datadog benefits from the greenfield opportunity of new cloud-native SaaS providers emerging everywhere, who may prefer Datadog over Splunk.
Those two factors allow Datadog to capture share from more legacy vendors like Splunk, but not entirely because of better execution in my view. Splunk just serves a more mature, lower growth market (though I suppose you could argue they failed to expand into higher-growth adjacencies nor were they able to manage the business for cash instead given they also have lower FCF margins vs. Datadog).
I agree there is not much real synergy between Datadog and Salesforce they do very different things. Splunk was clearly a better acquisition target and partner for Cisco than Datadog, for the reasons mentioned.
In my view, the only plausible M and A angle for Datadog would be a potential combination with some kind of next generation or so called neo cloud provider or I can see IBM/Oracle. But then again price is just too high for straight up acquisition.
The rest of your points sound reasonable to me. The key now is to think about where the SaaS market structure is heading. Do we end up with a fragmented landscape of many smaller players that rent cloud infrastructure, or does SaaS consolidate into a smaller number of large platforms that can justify building more of their own stack. Without a view on that, you cannot effectively size Datadog’s total addressable market. I also have no idea how big the mature and legacy TAM is and how much of it has been already captured(it would be good if you can chime on this). But, as I said I feel there is some level of stickiness so once customer is locked in, it is unlikely for it to switch out easily unlike the coding agent market of today where every week there is new launch in town and people alternate and switching costs are relatively low now. On top of that, there is still the open question of what direction Cisco’s leadership will take Splunk strategically.
I am going to give you a very simple example to think about that I saw yesterday. There is a unicorn startup called Gamma that makes slides with AI, and I also recently noticed GPT and Kimi make slides as well. Now question wether the players that own the foundation models are going to keep disrupting these smaller application layer companies. If so, the SaaS market will face a lot of pressure. I will add the caveat that OpenAI is a Datadog customer, but Datadog obviously benefits much more from having a large number of smaller startups and unicorns as customers too and they are also less likely to build their own cloud infra and tools. If this keeps happening that makes it challenging for Datadog to reliably grow its customer base.
@Mastery 7 IB Shitter
Thanks for responding and sorry for my delay. Was enjoying the last few days of Thanksgiving :) Hope you had a good break too!
Interesting point on neoclouds being a potential combination candidate, but how would that work? Is the idea that the neoclouds offer Datadog monitoring to customers as part of the overall package? I suppose neocloud customers renting infrastructure to build apps would be "cloud-native", so that sounds fitting for Datadog. This is where my technical knowledge reaches its limit though. Is that previous statement actually true? I don't really know.
---
Now on your second point about how the market structure evolves, it's a very interesting question! Certainly, there's a big question around what happens to the "application layer" as you put it. I'm going to define this as any software application that does not involve providing infrastructure hardware or software for compute, storage, and/or networking to the end customer (and for AI, I consider neoclouds and foundation model providers to be in this category as well, e.g., OpenAI).
I do think there will be some level of consolidation over time, but the tl;dr is I think there is a limit to consolidation. The first reason is cannibalization. At the end of the day, the application layer companies are customers or essentially commercial partners of the infrastructure providers.
For example, when Datadog's enterprise customers use Datadog's software to monitor their infrastructure deployed on AWS, they are sending data to Datadog's backend, hosted on AWS. Even if the customer isn't an AWS customer (say they have all applications running on Azure infra), they still drive traffic through AWS (and therefore generate revenue for AWS) by using Datadog. (Is that technically correct?)
Therefore, why would AWS spend any significant amount of investment on building out a competitive offering to Datadog for infrastructure monitoring? They do have a basic alternative for monitoring (AWS CloudWatch), but it's nowhere near Datadog levels of functionality.
Additionally, I think the issue with the infrastructure giants consolidating at the application layer is that a lot of the value proposition of the application layer is being infrastructure-agnostic. Sure customers can just use AWS CloudWatch, but part of why Datadog still has customers is that customers aren't exclusively AWS customers. They likely have workloads on multiple clouds, and that's where a vendor like Datadog can shine.
The other conideration is the selling motion and product engineering. AWS's selling motion is likely different from Datadog's. The former targets IT operations executives, whereas the latter targets application engineering/SRE executives. For some enterprises, those may be the same team but not all. That would require AWS to split their investment dollars between both across R&D and S&M to build out both go-to-market and product development organizations. That reduces the synergy opportunity.
I think the same logic applies to the AI model builders and the application layer built on top of those foundation models.
Now could there be consolidation among non-infrastructure application software giants? Of course, and we've already seen that happening with companies like Salesforce encroaching on customer support/contact center (vs. players like Genesys), marketing automation (vs. players like Klayvio), etc.
That said, consolidation is limited by the selling motion/engineering issue. There needs to be some level of adjacency to really allow for synergies to recoup the premium paid. Having advised many clients on the buyside, the key question when it comes to potential acquisitions (or organic initiatives to expand into new markets) is, "Do we have the right to play in this market?" Salesforce is unlikely to try and compete with or acquire cybersecurity companies for example because it requires a completely different sales and engineering skillset.
In terms of sizing the mature/legacy market and how much has been captured, a consulting person may be better equipped to answer that quantitatively. From what I've seen, growth in cloud computing is still expected to continue rapidly. It's unclear how much is driven purely by AI mania, but there are certainly still many mature industries that have yet to really shift the majority of worloads to the cloud.
It comes back to compliance and data privacy requirements. There's a reason so many operations are done on very dated technology in places like hospitals, banks, government agencies, etc. That's a big opportunity still as they migrate to the cloud over time, assuming the public cloud continues to develop security and data governance capabilities that fulfill compliance criteria. AI creates another tailwind, albeit a more uncertain one.
Given that it's my user name, figured I'd respond. I left IB in 2012 to found a health tech company software + services that went well, still around/growing/profitable, and am now CEO of a true software company in Fintech. I studied Mech E and Finance. I started coding around 8 years old and did a lot of code heavy work in Mech E for internships, side projects in college. I know how it all fits together, can comfortably talk to CTOs, but no one would hire me as a SWE at this point.
I love your story, and would love to achieve something similar one day. Any advice for a recently hired BB analyst who studied finance?
Entrepreneurship is hard and a big gamble. It is the same hours as IB but much more stressful as it is exclusively on you as the founder to make it happen. I like the search fund model, but sourcing is hard. I've had friends do very well with it and others who couldn't land a company, despite having financial backing.
I went from making >$750k in IB, depending on the year, in a tier 2/3 city and then took no salary for 2 years while investing most of my savings into making my first company work. People didn't really get it, but it worked. I ran my start up for 10 years and completely burned out. I also had a burning desire to build something and took on a massive challenge. I now own a ton of illiquid equity in a company I am no longer in charge of.
After I took a year off, I looked at some finance jobs in corp dev and VC. The recruiters all just wanted me to be a CEO again, and I didn't really want that. I joined a PE fund as an operating partner and asked too many questions on a struggling company and somehow ended up back as a CEO. First world problem, I know, but being a CEO is a hard, lonely role. I've spent the past 4 months since taking my new job fixing the culture, product and processes and buying another company, which just closed.
I'm not sure that's really advice, but it's been my path. I always wanted to create something from nothing and did so, but I think it also pigeon holed me a bit as I got a lot of pushback for non-CEO roles as to why wouldn't I want to be in charge or could I work for someone else, which I totally could. Am happy in my current spot and plan to be here for at least several years or until exit.
You mainly need enough technical understanding to model revenue and growth assumptions accurately, not to become a software engineer.
Autem consectetur expedita nisi nisi recusandae possimus error omnis. Eveniet autem in molestiae sit. Illum quos ipsam numquam enim quia corporis qui quas. Ea rerum non aut.
Adipisci harum aut maxime consequatur minima error labore est. Velit non quisquam dicta magni blanditiis dolorem. In suscipit iure eum qui voluptas. Molestiae et numquam labore quia.
Ducimus libero et sit optio est natus. Pariatur et nam id in aliquid. Ea blanditiis dicta aliquid natus. Qui suscipit molestiae necessitatibus voluptatum.
Rerum similique magni aut saepe aut eveniet tenetur non. Laudantium dignissimos dolore enim id aliquam. Vel sit enim deserunt. Quod ut voluptas omnis aliquid quos tempore perspiciatis.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...
Fugit quod sit enim quia. Sit qui dolorem est quis facilis voluptas omnis.