How Is AI Affecting Cybersecurity and Cloud Spending? Takeaways From the Cloudflare Earnings Report

Artificial Intelligence (AI) has garnered a fair amount of hype over the past two months. But investors may be focusing on the wrong set of companies. In this report we focus on two areas that we believe are overlooked and offer outsized opportunities: cybersecurity and cloud spending. 

Contents:

• AI consumer vs. industrial/B2B applications - where is the opportunity?
• What we learned form Cloudflare's (NET) F4Q22 earnings report
• Why we believe this tech re-bound is sustainable

AI consumer vs. industrial/B2B applications 

The consumer is generally quicker to adopt new technologies compared to enterprises. However, when it comes to capturing value it gets trickier as demand is not very sticky. If competition increases, the bulk of the value/benefit ultimately gets passed onto the consumer. 

An example is Microsoft introducing ChatGPT for Bing resulting in $100bn of market cap wiped out from Google. We expect Microsoft/Bing to gain some share in the near term, but there is no question that Google will step up and offer a competitive product. However, ultimately, a more efficient search engine is likely to result in less adds. We believe this will be the case for many other consumer applications and expect AI to be requirement rather than a differentiator.  

We are much more optimistic about AI for B2B applications and believe that AI has the potential to transform many industries (e.g., autonomous driving, robotics, life sciences, drug discovery), but we expect that it will take several years before we can see tangible benefits. 

Where we do see tangible benefits even today, is on the infrastructure side, ranging from hardware and software required to train AI models, to data storage, compute and cybersecurity solutions, required to scale workloads. We have written extensively about Nvidia on the hardware and software side. In this report, we will provide several use cases on the cloud and cybersecurity side.  

Cloud and cyber AI use cases - Cloudflare F4Q22 takeaways 

Two areas that we rarely hear in the context of AI are (1) cloud storage/compute and (2) cybersecurity. We believe investors are not appreciating (1) the data requirements associated with AI (2) the incremental cyber vulnerabilities that will get created as AI use cases increase. 

Here are few examples: 

• Many AI processes will require a more dynamic use of data vs. data stored at one cloud with high egress fees (i.e. costs to take the data out). Many will require compute power closer to the source (edge)  
• Increased edge use adds incremental vulnerabilities; increased web traffic requires more bot management, CDN, DDoS services 
• AI/Chart GPT just made e-mail phishing a lot more effective. E-mail is a major attack vector and a common entry point for cyber breaches

These examples are already finding real world applications. Below are use cases/ customer highlights from the Cloudflare's F4Q22 :

• The largest current R2 customer is an AI company that needed to find ways to cost efficiently run their models across multiple data providers. "They are by nature, multi-cloud, but the data egress policies, make it prohibitive to move large training set between the clouds" and "Cloudflare's Workers and R2 become the glue". With R2, companies can use their data in a dynamic way as it is essentially "stored on the network".
• A leading generative AI company (a.k.a. OpenAI/ChatGPT) signed a $1mm deal with Cloudflare after the company saw traffic explode in November and needed CDN, DDoS, Bot management, Gateway DNS etc.
• With increased phishing activity, we expect more demand for Cloudflare's e-mail security product (Area 1). According to Cloudflare, adding e-mail security for its large installed base of Application Services customers is a simple DNS change (1 click install). This can serve as a gateway for Cloudflare to access more Zero Trust customers. 

As background, Cloudflare is a leading content delivery network (CDN) and DDoS mitigation company (acts as a reverse proxy between a website’s visitor and the customer’s hosting provider). The company has built a network of 270 points of presence (POPs) across more than 100 countries globally offering a low-latency network. Cloudflare operates at a massive scale with over 4 million non-paying customers and 151,000 paying customers.  

Cloudflare has a unique business model where the main product is Application Services (CDN & DDoS mitigation), but the company can use its current infrastructure and customer base to expand into other areas (with limited incremental investment) such as Zero Trust and Network Services (SD-WAN) and compete with leading vendors such as Zscaler and Palo Alto Networks; and Workers + R2 Object Storage + D1 Database where it will compete with the hyperscalers such as AWS, Azure, GCP.

While Zero Trust, Workers, R2 Object Storage are still in the early innings (~10-15% attach rates), it is interesting that Cloudflare can do all this using its existing infrastructure. The company's utilization is at ~75%, with meaningful excess capacity that can accommodate the announced product offerings without a significant incremental investment.

Cloudflare operates as a network with a “scheduling engine” vs. “server rental” which allows the company to run close to capacity, compared to AWS utilization in the high teens, highlighting a key point of capital efficiency. The company employs JIT infrastructure allowing it to double the size of a major location in 30 days if needed.

Why we believe this tech rebound is sustainable

While most management teams are pointing to weak demand, extended sales cycles, difficulty in acquiring new logos, etc., it is important to point out that enterprise spend is a backward looking indicator. The areas that we track as forward looking are semiconductors, discretionary spending and commodities, all of which are showing signs of bottoming. 

A typical sell off follows a pattern where risky assets sell of first, which is than followed by a broad valuation re-set, and lastly, earnings estimates get cut. Technology, especially riskier tech companies got disproportionately hit in the first two phases of this downturn. For SaaS companies, for example, valuation re-set from >20x to ~5x EV/NTM Revenues.

Investors often misunderstand the impact of interest rates on technology and are under the belief that technology investments can not work in a high interest rate environment. This is not correct. While interest rates are an input in our valuation framework, once rates settle in a range, the stock performance is driven by earnings upside/downside. As a sensitivity, 1% increase in interest rates results in ~15% downside for most companies under our coverage universe, a one time impact.

We hosted a (free) webinar in November 2022 where we covered (1) interest rate impact on valuations, (2) the three phases of the downturn and (3) where we are finding opportunities. The webinar is now available on demand and eligible for 1CE CFP credit. Reach out if you would like the replay access details.  

While there may be still a lot of volatility ahead, we believe that an investment opportunity like this only happens a few times in a century. 

The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.

In This Story

NET

Other Topics

Cybersecurity Cloud Computing Technology Earnings