I have been catching up with Keith Ashworth-Lord’s (KAL) Buffettology factsheets. This extract from the June update caught my attention:
“On 30th April, we invested in the IPO of Darktrace where dealings went unconditional on 6th May. This is a business that I have known and watched for several years now, having become acquainted with it by virtue of research on the market for the assurance division of NCC. It is the world’s first ‘cyber AI’ company, having been founded in Cambridge 2013 by the combination of cyber experts working on national security with mathematicians with an interest in machine learning. The company’s AI Immune System Approach is self-learning, providing a surgical response to a breach at machine speed. This is the key differentiator and where the IP and economic moat lies. The analogy here is the human immune system which learns what is normal and what is abnormal and acts accordingly. A notable client list counts Siemens, Vodafone, Coca-Cola among 4,500+ customers in 110 countries, with North America the main overseas market. The road to profitability in the medium-term is, we contend, based on projected sales growth of around 30-35% p.a. and mid 20% operating margins. Apart from the current financials – which we are confident will rectify in time – Darktrace ticks all the boxes of our Business Perspective Investing checklist.”
A few points:
I did not know cyber AI fell within KAL’s circle of competence.
Can this eight-year old business have already created a demonstrable moat? (I guess the history of Google shows this could be possible.) DARK claims its competitive strengths are:
Leading AI platform in enterprise computing with first application in cyber security
At the forefront of the next evolutionary wave in the cyber security industry
The leading autonomous cyber security AI products, with ability autonomously to detect, respond and investigate threats.
Differentiated sales model and strong value proposition for customers and partners drive rapid new customer growth and long-term customer relationships
Broader adoption of technology platform enables substantial growth opportunities
Attractive financial model combining rapid growth at scale, revenue visibility, strong underlying profitability, and cash generation potential
Experienced management team with strong track record of execution and deep AI expertise
Trailing revenue was $234m before the flotation, which, if compounded at 30% for, say, five years, would give $870m. A 25% margin would then give an operating profit of $218m, or £157m at GBP:USD 1.39.
DARK’s offer price was 250p, which supported a £1.7b market cap at the flotation.
Perhaps KAL thought paying a pre-tax profit multiple of 11 for 2026 was good value. I wonder if he has watched this interview with Mr Buffett, during which the old master gave some IPO valuation advice. Mr Buffett explains if a flotation is valued at $50b, you should certainly be expecting it to be earning $5b pre-tax after five years.
The best buy price retail investors could have achieved was 312p a week after the flotation. So buying at 13-14x pre-tax multiple for 2026.
"Darktrace now expects a year-over year increase in its constant currency ARR of between 32% and 34% (previously 26.5% to 29.5%). It also now expects year-over-year revenue growth of between 29% and 32% (previously 27% to 30%), considering normal quarterly seasonality patterns and particularly, typically soft first-quarter sales. "
I must admit DARK would not seem to be a textbook Buffettology purchase. Back in the day KAL was buying Liontrust and Games Workshop, both smaller businesses he was quite familiar with, and both became mega-baggers after being turned around by fresh management.
Even after the shares have almost tripled, DARK does not feature within Buffettology’s top ten holdings. I get the impression the investment was relatively small for the fund.
Nonetheless DARK does show KAL is willing to invest in sectors and at valuations that traditional Buffettologists may feel uncomfortable. Whether buying DARK leads to KAL buying further ‘rich’ IPO valuations remains to be seen.
I would welcome any insight/comments on DARK or KAL’s decision to invest.
I’m ashamed to say I purchased a small amount based just on KAL’s recommendation, and the sector involved. The thing almost bagged and I sold half effectively leaving myself in for free. I deserved to lose money for listening to someone else and not doing my own research, however I intend to ‘coffee can’ my current holding for the long term. Totally undeserved!
There is a US legal issue with ex DARK director, Mike Lynch, over his sale of Autonomy to Hewlett Packard (the US say there were accounting irregularities in Autonomy’s books- I have no idea who is right). Mr Lynch currently owns 4.4% of Darktrace and this could draw the DARK into US money laundering legal issues (presumably because the Autonomy sale proceeds may have been invested in DARK), with Mr Lynch being extradited to the US. There was a disclosure of this in the DARK admission document. A Court ruling in favour of Mr Lynch’s extradition may have caused the recent sp weakness, but he can still appeal this.
My take is that KAL may think this is a platform / network business similar to RMV or Softcat, where revenue growth of 30%+ will eventually lead to huge EPS increases.
Well, KAL is a good source of ideas but perhaps not the most reliable approach to adopt! Good point on Mike Lynch. The flotation document is full of risks related to the Autonomy sale, although I would not say they would directly impact DARK’s ability to develop and sell its software. Mr Lynch’s wife owned 12% of DARK at the float, too. (Not sure if the Lynchs have made more money on DARK than Autonomy.)
Good point on the network effect. I wonder if DARK’s software improves in proportion to the number of clients involved i.e. – a virus attack ‘learnt’ on one client site can then be applied to all client sites. Once you have a critical mass of clients that has generated significant AI ‘learning’, a copycat could find challenging DARK difficult. WordPress security plugins (e.g. Wordfence) use a similar ‘network’ approach, although the coding is somewhat less sophisticated than DARK’s!
thanks for these notes Maynard. DARK is a co that I have on a list to minitor for now, as I find the current price not attractive enough to make an investment, and as with most IPOs I expect there somewhere along the near future will be some kind of unexpected or untoward event that will create a more preferable chance for me to take a starter stake.
I did take some time to check their website, specifically careers and they had a multitude of engineering and matematician roles available, so I really got the sense of a confint company with plans to go places and execute.
The toruble I find with these kind of stocks is, as you say, can we ever undertstand enought, truly, to determine how much of a competitive moat they command. or a bit like google, do we have to take it in fath and let the future prove that for us. It is certianly a refreshing approach to the cyber problem which until now has seemed to be a case of ‘check all the doors are shut and store the key locks somewhere safe’ now we are talking proactive monitoring of large corporation estates. that is quite a position of trust, and assuming it works then could generate them multi year tie in revenues from customers too scared to leave…
Firstly, the Lynch/ Autonomy thing is a total non event for Darktrace. It’s the kind of thing the lawyers can get worked up about and put risk factors in the prospectus, but it’s just irrelevant to a Darktrace shareholder (although not to Mr Lynch, obviously).
I’ve watched KAL for a while. He’s a very interesting investor, and Dark is a very interesting company. On KAL first of all, it’s a bit much calling yourself a Buffettologist and then buy into richly valued IPOs of companies that aren’t even profitable yet! I agree with Maynard that this seems quite far removed from what made him successful. I’d question a little bit whether he really understands this business (the competitive dynamics in particular). His problem is twofold as I see it. Firstly his fund has got big and so he has to buy big companies now, even though some of his big successes were smaller companies like GAW that had potential that then exploded. It’s possible to do well in big companies obviously, but it’s harder - and he’s not been totally honest about this in my view.
Secondly I wonder whether he’s suffered a bit of style drift. Darktrace and other purchases, like Rollins, have been very very expensive. Rollins for example is an excellent company but I reckon he bought it on a 40x p/e or something like that. That’s a hell of a valuation headwind and more importantly, it doesn’t leave any margin of safety. It’s a strategy that has worked recently, though, but for me it’s a bit of a failure if a professional can’t find a few investments of equal quality at better prices than that. To be fair, this is certainly harder if you have to look at big companies only. And at least with Rollins you’re certainly talking about a quality company. I wasn’t convinced by his purchase of Berkshire either. Berkshire isn’t the business it used to be - it’s got a fair few capital intensive companies now, like BNSF, the utilities and suchlike - and I see it suffering poorer returns in future. If it does a lot of buybacks and maybe starts paying a dividend it could do well though.
On DARK - it’s the kind of thing that could do very well. You could imagine clients feeling really quite locked in once they have subscribed to DARK’s services, and high customer retention is usually the sign of a good or even great business. However, it’s difficult to assess growth, margins and capital allocation, so it’s tricky to come to a view.
I’m interested from which stand-point are you saying there is no risk from the ‘Lynch/Autonomy thing’?
Are you a corporate lawyer, or is it from past experience that you say this and if so, which experience? Or is it just a feeling that you have?
I am no lawyer, but if the company was set up with fraudulent funds I would think under money laundering regulations Darktrace could be banned from trading with US companies? Why would this not be one possible outcome?
Pretty sure KAL has bought more DARK. Latest Buffettology factsheet (for the month of August Buffettology Fund Fact Sheet 2021.9.pdf (806.1 KB)) shows DARK in 8th position at 3.14%, whereas previous factsheets never disclosed DARK as a top 10 position. Rough sums based on month-end share prices suggest DARK is now a £57m investment with c9.2m shares versus a maximum £51m investment with c6.6m shares at the end of July.
“Self-Learning AI is made up ofmany thousands of algorithmsthat inform its decision-making, each with different strengths. These algorithms operate in competition with one another to deliver the best model for every user and device.”
"To determine which algorithms to employ at any given moment, Darktrace uses asmart threshold filterthat weights and re-scores the outputs from all machine learning detectors in light of their previous performance."
Here is the complex maths bit.
"To combine these analyses of digital activity, Darktrace uses a technique known asRecursive Bayesian Estimation. Crucially, this allows the AI to continually recalculate threat levels in light of new data and discern significant patterns in data flows indicative of attacks."
How Recursive Bayesian Estimation works with DARK:
Different data from different devices employed by different users is fed into different algorithms;
Over time the algorithms will exhibit different strengths and weaknesses when differentiating between ‘normal behaviour’ and suspicious activity for each user/device;
DARK uses a “unique” Recursive Bayesian Estimation model, which applies ‘probability theory’ to determine the most accurate algorithm to assess any suspicious activity.
In other words, the algorithms are effectively competing with one another to deliver the best results. This effective ‘classification’ of the algorithms is what DARK says represents “one of its key inventions”
The algorithms are also enhanced through ‘expert human behaviour’:
DARK automatically tracks every cyber-analyst employee on their computers;
This tracking ‘learns’ what these employees do (i.e. how a threat was interpreted, the subsequent action etc), and;
The details are fed back into algorithm development.
Anybody familiar with Recursive Bayesian Estimation?!
Q&A 39m15s: DARK’s product manager on competition and rivals Vectra and Extrahop:
“We rarely encounter competition in deals, we very rarely lose to competition and we just do not see these organisations catching up. They are very much still ‘point products’**. They don’t have the breadth we have in terms of being able to operate in the email space, across all types of cloud, not just a certain vendor, the industrial environment can’t be supported by those technologies, and these things are all a testament to the fact that their technologies aren’t able to learn on the job and be broadly applicable in customer environment. When you add on top the lack of autonomous response, so no ability for these products to react to attacks that they don’t know about and interrupt the incident before it turns into a crisis, we find it very easy to differentiate ourselves from them and maintain the prices we want to sell our product at.”
(**point products generally protect a company against only specific and common threats)
Darktrace is certainly a challenge to understand and a greater challenge to judge if it is investable at this time.
I have had some exposure to developing AI-based software in my career but this field moves so fast that the modern set of software tools and architecture leaves me at the gate with a mouth ajar.
Their prospectus goes some way in explaining the science behind their products. Frankly, in my days, we used to ask, is this real or is it vapourware? To be honest, I can’t say one way or another.
A disappointing, and possibly a red light warning is the fact that company has not said anything about the competition landscape in their prospectus. If they claim their product has several unique features and a novel approach, what better than highlight this with a competitor analysis?
I googled to dig deeper and it appears that they do have several competitors who also claim their products are superior to Darktrace. Several of them are US public companies with much larger market caps, and established a lot longer.
Now to fundamentals. Sharepad conveniently provides broker forecasts looking forward 3 years. These show the company to be loss making all the way with little sign of the losses reducing. My back-of-fag-packet calculations show that they will have to at least double sales each year for the next 3 years just to break even.
Which brings me back to KAL investing in this company I think from the start. KAL must have looked and analysed the history of this company and forecasts, as I have done.
So what metrics or rationale is he using to invest? Not based on fundamentals, I would argue.
The only explanation I can come up with is a belief that he is acting as a trader. Somehow he has judged that this share will fly based purely on demand and supply, in the current ‘sexy’ world of AI security software and the need for effective cybersecurity. He will probably wait till it multi-bags several times and then dump it.
DARK still KAL’s fourth largest holding at 4.4% and £75m value.
Difficult to read whether KAL has bought more, as DARK’s month-end price was 802p but a few days later when KAL’s commentary was dated the price was 600p. Taking the 802p does not suggest a major top-up, but 600p of course would.
"As regards Darktrace, where aspersions have been cast about the quality of its R&D and product offering, I will leave you with this thought.
Bytes Technology is a software and hardware reseller than we own in Free Spirit. On its results call at the end of the month, Neil Murphy (CEO) said, “I know Darktrace has been in the press recently, but every single one of our Darktrace customers is renewing and customer satisfaction is very high”.
Bytes is Darktrace’s EMA Partner of the Year, so that’s a material sample size of customers that he is talking about. This type of price action is precisely the sort of stock market noise that I always counsel against listening. I do so again."
With the shares now at c600p, could be an opportunity to follow KAL’s DARK top-ups from August.
I took a small stake a few weeks back when the SP dipped just below 600.
I have now formed the view that either this company’s products are genuinely a step up from the competitors’, or the whole thing is plain hype. If it is the former, then it is massively undervalued. If the latter, then Peel Hunt are right, and it is worth in the 400’s.
Keith Ashworth-Lord’s commentary in the November Buffetology Fund Factsheet includes the following snippet;
‘As regards Darktrace, where aspersions have been cast about the quality of its R&D and product offering, I will leave you with this thought. Bytes Technology is a software and hardware reseller that we own in Free Spirit. On its results call at the end of the month, Neil Murphy (CEO) said, “I know Darktrace has been in the press recently, but every single one of our Darktrace customers is renewing and customer satisfaction is very high”. Bytes is Darktrace’s EMA Partner of the Year so that’s a material sample size of customers that he is talking about. This type of price action is precisely the sort of stock market noise that I always counsel against listening to. I do so again.’
I don’t currently hold Darktrace but it’s beginning to look more interesting.
Hi Snazzy - yes, things are looking very interesting, perhaps even in dangerous territory.
The current sp is now 516. It has beeen sliding ever since PH’s announcement. At this rate, it might sink below 500 by the end of this week, who knows.
If KAL is right, I should be topping up. But the other side of my brain is saying don’t. Why is the sp sliding when the company keeps announcing great news of more contract wins? Hhhmmm.
I grabbed a few today at 500p more just to keep an eye on it, and sufficient that if it didn plunge 50% on news that would allow me to exit or to take advantage and buy up.
It is tricky for me to buy a company valued in excess of £3bn with no profits projects for the next couple of years. But I too am loathe to miss a potential opportunity right now, so ‘dipping my toe’ allows me to be invested and to keep a close eye on what is happening, and not miss it if it rapidly ascends from here.
I do not think this field of cyber sec is KALs speciality, for me it feels a bit like Neil Woodford getting caught up in bio techs…
For me, Peter Bonfield stands out. He was the guy in charge during the slow decline of ICL in the early 90s. The guy who had to fly to Japan to tell Fujitsu that their new acquisition had used all the accounting tricks available to juice their numbers and that there was not much left to exploit. He then took charge of BT in the late 90s when the shareprice rose from £4 to £15 and then back down to £5. This was during the period when BT was desperate to link up with a US Telco and got fatally involved with MCI and then AT&T, and also decided to go hell for leather expanding its mobile networks in Europe. The US deals failed and in bidding for mobile spectrum BT took on so much debt that it nearly went bankrupt. Lots of lessons for Mr Bonfield!
However, it looks like the directors don’t have a lot of skin in the Dark Trace game.
"I was delighted that on the 6th May our dealings into the Darktrace IPO went unconditional. I have followed this business off market for several years and it entered the portfolio at an attractive valuation.
Darktrace is all that is good about the UK tech sector and is a true world leader in cyber security. Its economic moat is an AI Immune System approach that provides a surgical response to a breach at machine speed.
The Deliveroo debacle increased the need for Darktrace to go well and there were also some red herrings; Darktrace’s profitability and a former non-exec board member facing criminal charges in the US. The reason for the former was music to my ears,they invest all profit back into the business particularly sales & marketing. They are cash flow positive and could be profitable tomorrow if they chose to be. The latter is irrelevant to the day-to-day running of the business and its processes.
This confluence of factors meant the IPO was priced to go and Darktrace has achieved a 100% increase from my initial purchase price quicker than any business I have ever invested in."
Perhaps KAL has adjusted DARK’s P&L for sales and marketing costs. This from the 2021 accounts: