Why I just sold Zscaler!

Hi everyone, and welcome new posters!

As the title says, I just completely sold out of Zscaler. For context, this was a 1.8% position in my portfolio. After working through the latest earnings report I realized that the investment thesis has significantly changed. The signs of this already started showing up in the previous two earnings reports, but this one made it crystal clear:

Zscaler’s Capital Efficiency (CapEx vs FCF) collapsed in Q4 as they reported a huge CapEx spike of $199.8 million, which represents a staggering 22% of revenue for the single quarter. Management tried to pitch this as an “opportunistic” pull-forward of data center infrastructure to secure hardware pricing and brace for AI workloads. That might be true, but if we look at the trend over the last year, this isn’t a one-off anomaly; it’s a steady, multi-quarter deterioration of capital efficiency. Look at how Free Cash Flow (FCF) margins have steadily fallen year-over-year:

  • Q2: FCF margin dropped to 20.7% (down from 22.1% YoY)
  • Q3: FCF margin dropped to 16.0% (down from 17.6% YoY)
  • Q4: FCF margin collapsed to just 6.8% (down from 23.9% YoY)

Again, I am not saying here that increasing Capex as they did is necessarily a bad decision - it might in fact be the right decision for the business to survive and maybe even do well in the future. But one part of my core SaaS thesis for Zscaler relied on the business being asset-light with high operating leverage. Zscaler is starting to look like a heavy infrastructure company just to keep its platform relevant.

This by itself wouldn’t necessarily be a deal-breaker for me, but combining this with the growing disconnect between CapEx and slowing ARR growth/revenue growth is what ultimately made me sell this position for now. If management truly believed these huge upfront hardware outlays would fuel hyper-growth, it should show up in the forward guidance. It didn’t: For FY27, management guided for just 16.6% to 17.4% YoY ARR growth. Compare that to their initial FY26 ARR guidance, which sat at 22.7%.

So, we are being forced to accept much heavier capital requirements and lower cash flow margins for a business whose top-line growth is rapidly decelerating. Even just looking one quarter ahead, I am now penciling in that YoY revenue growth will decelerate to 22%, from 24% in the just reported Q4 (mid-point Q1 revenue guide was just for 18.9%).

Digging into the backlog also reveals something interesting: While current RPO growth held steady at 24.7% YoY (even up slightly from 24.4% last quarter), Total RPO growth has now decelerated for three consecutive quarters. Total RPO YoY growth has fallen from 34.5% in Q1 to 27.4% in Q4. When total RPO drops faster than cRPO, it means customers are still renewing short-term, but they are increasingly pushing back on signing large, multi-year expansion contracts.

I don’t like seeing a premium SaaS company transition into a capital-heavy defensive game when growth decelerates from 25% to likely ~20% for the upcoming FY, while CapEx climbs exponentially and long-term customer commitments slow down. With that there is a good chance that the unit economics might be fundamentally broken. I say “might”, because I can also see a bull-case where everything works out wonderfully with the increasing demand for security in a world of ever evolving AI threats. But as my friend @stocknovice likes to say, we are trying to make 60-40 bets with our investments and right now Zscaler feels more like 40-60 to me.

So, I chose to put the capital elsewhere. Where, you might ask? I split the proceeds between Astera Labs and Axon. Both had outstanding earnings reports as you can read here and here and I feel my higher confidence positions already have high enough portfolio allocations. I will say that both Astera Labs and Axon aren’t free of risk either:

For example, one thing that bothers me with Axon is management’s goal to build a flashy, sci-fi-inspired $1.3b headquarters campus in Scottsdale, which I think is completely unreasonable. I mean, does a public safety tech company really need to develop over a thousand residential apartments and a hotel as part of its corporate footprint? It feels like a distraction and an unnecessary use of capital when they should just focus on their core business. Yet, the rest of the Axon story is just too good to not invest in. Despite the real estate distractions, the underlying business is essentially a hyper-growth software powerhouse disguised as a hardware company. They have locked down a near-monopoly on law enforcement technology, and their transition into high-margin cloud ecosystem subscriptions has given them an incredibly sticky high-growth revenue model that is very difficult for competitors to disrupt.

As for Astera Labs, I am a bit concerned about long-term visibility of the business fundamentals because things in the AI infrastructure world can change quickly and hardware standards are constantly in flux. While their execution right now is flawless; evidenced by Q2 revenue hitting $392.4m and my projected Q3 acceleration to 156% YoY growth when assuming a typical 7.5% execution beat - their revenue relies heavily on preventing data bottlenecks between other companies’ chips. In the AI hardware landscape, if a major hyperscaler customer shifts from PCIe architectures to proprietary optical fabrics, or decides to vertically integrate connectivity directly into their next-generation custom silicon, Astera’s hyper-growth could face sudden headwinds. This underlying structural risk might be exactly why the stock hasn’t moved up an inch in the weeks after their mind-blowing earnings print and next quarter outlook; I think the market is seeing huge near-term growth numbers but treats the long-term durability of that revenue with a healthy dose of skepticism.

So back to Zscaler, I’d love to hear if anyone else is seeing these same red flags in the backlog, Capex investments and/or growth deceleration, or if you are choosing to trust management’s long-term AI thesis and why. And happy to hear anyone’s thoughts on Axon or Astera as well!

Best,
Ben

47 Likes

Hi Ben

Nice to swap notes. Actually - yes I feel the same way about ZS.

I’ve been holding 4-5 cybersecurity shares in my portfolio for a while (ZS, Crowdstrike, Rubrik and Sentinel One - plus Cloudflare depending on how you categorise them) and always thought for the whole time that SentinelOne was the ballast I’d been carrying, (mostly due to poor share price performance).

Looking at it again this month, I realised it was ZS. They are a leg down in their organic growth rate (revenues, ARR and backlog/RPO), vs their peer cohort once you take out their acquisition juicer, failing to seize the market with their zero trust value proposition, hated by users* and way behind on their respective ratings and share price valuation.

I had already sold Sentinel One into strength last month but this week trimmed and then cut ZS. I thought the investor day was totally uninspiring, the organic BAU numbers were pathetic, (5-10% below their peers), and the aspirational extra 2bn in revenues that they imagine agentic AI might deliver them on top of the $8bn ARR target for 2031 just laughable - which analysts basically admitted afterwards in their coverage.

I withdrew the trimmings and re-invested the rest into NetSkope which is expecting to grow 50% faster and not outrageously valued but faring significantly better in its progress towards its own 52 week high. I’m happy to stay exposed to Cyber Security but only in the right players.

Ant

*I had to use ZS at a client’s offices last year in Singapore. The entire process of sourcing a freaking wifi connection to present to the client was a total nightmare, involving a physical workstation and remote tech support, a multi step multi hour process - just to get a wifi login with no access to their network.

17 Likes

Ben,

I have traded in and out of Zscalar (ZS) since 2018, most recently selling my position in January this year.

My reasoning at the time was that I was not satisfied with their approach to the problem of AI security risks. This was a subjective analysis from reading ZS earnings call transcripts and comparing what was said to other companies.

Much of this comparison comes from Crowdstrike (CRWD). My interpretation of comments by both companies as of early this year was that Crowdstrike was approaching AI-related capabilities and challenges proactively as an opportunity while Zscalar was reacting with no clear vision. I had no faith Zscalar would thrive with the upcoming challenges related to AI.

Specific to the most recent earnings, all the warning signs you mention only confirm my early holistic analysis. I would not invest in Zscalar at this time. From a more general perspective, I expect the company would be a market-beating investment. Yet I have no faith that in the near future they would offer returns we look for in a Saul-style portfolio.

That said, Zscalar does remain on my watchlist in case their market position improves.

9 Likes

Nice write up for ZScaler! It’s interesting how the contrast with Cloudflare has played out over the years. I always considered Cloudflare to have the superior product and engineering. However, ZScaler had a much stronger go to market motion, better enterprise sales channels, and got FedRamp certified sooner.

I believe the differences in product come down to Cloudflare engineering its SASE solution to be Cloud native, whereas ZScaler was combining a number of legacy solutions. Just anecdotally, I’ve seen engineers have a good opinion of Cloudflare and heard a lot of negative things about ZScaler, especially about being slow.

ZScaler really did a phenomenal job with enterprise sales in the early days of SaaS. They worked extensively with channel partners and had all the right certifications like FedRamp for government contracts. It seemed like Cloudflare was playing catch up, having to re-org their sales, and possibly too much focus on engineering only. It has taken some time for the superior products to play out in the marketplace, but I’m guessing this is why Cloudflare’s trajectory right now is better.

From my perspective these two companies are practically a case study on why both strong R&D and strong S&M are important for a business. Strength in one area may compensate for awhile, but longer term both innovation and the ability to close sales effectively are important.

18 Likes

For example, one thing that bothers me with Axon is management’s goal to build a flashy, sci-fi-inspired $1.3b headquarters campus in Scottsdale, which I think is completely unreasonable. I mean, does a public safety tech company really need to develop over a thousand residential apartments and a hotel as part of its corporate footprint? It feels like a distraction and an unnecessary use of capital when they should just focus on their core business.

It is odd that they want to get into the hotel business. They believe the apartments are necessary to attract engineering talent and create a work/life environment to bring talented engineers from California into Arizona. This will be a 400k square foot office facility as well as manufacturing so it’s not just office and apartments. Axon currently has 900k square feet of manufacturing/office in Arizona as it is now that they believe they’ve outgrown.

Beyond its proposed $1.3 billion Scottsdale campus, Axon has aggressively expanded through more than $2 billion in acquisitions involving AI-powered 911 systems, drone detection, surveillance integration, and robotics. It’s also investing heavily in manufacturing expansion, R&D, and new products, while projecting approximately $600 million in annual stock-based compensation. Although these investments could strengthen Axon’s competitive moat, investors have become increasingly concerned about excessive spending, shareholder dilution, declining margins, and weak free cash flow. The company’s September announcement of roughly $1 billion in convertible debt further intensified those concerns, contributing to a sharp stock decline despite revenue growth exceeding 30%

You are buying Axon at a time of pessimism due to these investments. Axon continues to grow 30%+ and they do have a strong moat/ecosystem but these capital expenditures have occurred at a time where the stock market is focused on near term profits.

To me this is simply investing for growth, and, while I don’t like them getting in the hotel business, I do not see it as a thesis breaker. What has me more concerned is, for really the first time, Axon may have a formidable competitor. For years Motorola was really their only signifiant competitive threat in the camera/evidence tracking business. But now, Axon faces a similar crime tech competitor, in Flock Security. Flock and Axon were previously partners but as they began encroaching each other’s space, they parted ways. Axon is developing products that compete with Flock and Vice versa. What this means is if Axon has the relationship with the police departments, where they can cross sell, Flock is now developing that same business relationships where they. can cross sell competing products to these departments as they spread beyond security cameras.

Flock is moving well beyond license-plate recognition cameras toward becoming a comprehensive public-safety technology platform.

  1. Flock Alpha Drones: Autonomous drones dispatched to emergencies, competing directly with Axon’s drone strategy.
  2. FlockOS: Centralized software platform integrating cameras, alerts, investigations, and real-time operations, competing with Axon’s software ecosystem.
  3. Flock911: Emergency call and dispatch intelligence, overlapping with Axon’s Prepared and Carbyne acquisitions.
  4. Flock Nova: Investigative software combining data from multiple sources, competing with Axon’s investigative tools.
  5. AI Video Search: Natural-language searches across surveillance footage, competing with Axon’s AI-powered investigative capabilities.
  6. Mobile and Fixed Cameras: Expanded surveillance cameras, mobile license-plate scanners, and surveillance trailers, competing with Axon’s growing camera portfolio.

In August 2026, Flock CEO Garrett Langley said that license-plate recognition accounted for less than half of the company’s forward-looking revenue, and that drones were probably its fastest-growing business. Axon explicitly identifies Flock Safety’s Aerodome as a competitor to Axon Air, and Flock’s acquisition of Uniform Sierra Aerospace as competition for Axon’s Sky-Hero tactical drones.

20 Likes