the following comment in the article is not correct: ‘Each £1 of HGT shares is earning nearly 30p per year’. The HGT portfolio valuation is £3.2bn which at a 24x EBITDA valuation implies an EBITDA of around £130 million per year. The market cap of HGT is £1.7bn, therefore this implies an EBITDA yield of around 8% and an earnings yield of likely between 4 and 6% per annum (and not 30%).
the following comment about the following comment in the article is not correct: Each £1 of HGT shares is earning nearly 30p per year’. The HGT portfolio valuation is £3.2bn which at a 24x EBITDA valuation implies an EBITDA of around £130 million per year. The market cap of HGT is £1.7bn, therefore this implies an EBITDA yield of around 8% and an earnings yield of likely between 4 and 6% per annum (and not 30%).
24X is how many times its earnings people are prepared to pay to buy a business (aka the exit price).
34% EBITDA margin is how much, on average, each of the holdings earn per year.
One’s a profit figure, the other is a purchase figure.
But the statement ‘each £1 of HGT shares is earning nearly 30p per year’ is wrong. The EBITDA Margin of the underlying business is indeed 34%, but the HGT portfolio generates around £130m of EBITDA per year (£3.2bn portfolio valuation valued at 24x means circa £130m EBiTDA). The market cap of HGT is £1.7bn which means that the portfolio EBiTDA as a percentage of market cap is circa 8%. Hence every £1 of HGT shares earns circa 8p of EBITDA per year and probably between 4 and 6p in net income.
The sentence was referring to EBITDA which means before interest, tax, depreciation and amortisation and I used the word "EBITDA" numerous times before saying "each £1 of HGT shares is earning nearly 30p per year".
But I take your point that it could be misconstrued so I've revised the sentence to say "each £1 of HGT shares is earning nearly 30p of EBITDA per year" to provide better clarity.
It is not correct, £1 of HGT shares is not earning 30p in EBITDA each year. The £156m in trading result in the valuation bridge table is not the quarterly or semi-annual EBITDA produced by the portfolio, it is the increase in portfolio valuation as a result of increased portfolio EBITDA times the 24x EV/EBITDA valuation multiple which means an approx £6-7m increase in portfolio EBITDA.
The HGT portfolio is valued at £3.2bn which at an extremely high 24x valuation multiple means that it is producing circa £130m in EBITDA per year. When you divide £130m by the HGT market cap of £1.7bn, you get a circa 8% EBITDA yield which means that £1 of HGT shares produces 8p in EBITDA per year.
Trading EBITDA is not £154m per quarter. The £154m is the increase in portfolio valuation as a result of trading. To get to the £154m increase in valuation you need an increase in portfolio EBITDA by £6-7m which at 24x EV / EBITDA valuation results in £154m valuation increase in portfolio valuation.
The annual EBITDA of the HGT portfolio is between £130 and £140m per year which is approx £33-35m per quarter.
Please don’t make any investment decisions based on the assumption that every £1 of HGT share price earns 30p of annual EBITDA as the reality is that it is (in my estimation) around 8p per £1 of HGT shares.
Also,,by referencing it to earning per £1 of HGT shares you bring valuation into the equation.
This. The OB is mixing up the way they report earnings growth with earnings.
Also: they report EBITDA growth INCLUDING M&A; it's not organic growth, so careful with Rule of 40.
Separately (and missed from the analysis above): their average business had a crazy 7.1x net debt/EBITDA at the end of 2021. I'm REALLY surprised more didn't go bust/equity valuations didn't crater when rates rose
No mix up. I was referring to EBITDA earnings but used the word "earnings" loosely and have added the word "EBITDA" for better clarity.
Thanks for raising the point about the level of debt from 5 years ago which you describe as "crazy". In that year HGT raised $125m junior bond financing to support its portfolio. That ratio went even "crazier" in 2022 hitting 8X and today it's above 2021 levels at 7.2X. I'm struggling to understand the point you're trying to make about the relevance of the 2021 ratio of net debt to EBITDA.
I don’t know why you want to get involved here. Sentiment around software is so bad due to (a) uncertainty around terminal value, which will not be resolved in the short term (b) requirement to accelerate investment in AI capabilities that will reduce margins in the near term (c) current software multiples (24x for HGT) still do not reflect a particular pessimistic survival rate suggesting room for further downside. Just my 2 cents
Teubsch, did you even read the article? I don't know why you don't know if you did read it.
To your points
a. "uncertainty around terminal value, which will not be resolved in the short term" - When was there ever certainty? The sentence is meaningless as you're basically saying crystal balls are not about to be invented and the future is uncertain. Certainly is!
b. Requirement to accelerate investment. Yes, HGT have accelerated AI investment and their margins have grown (read the article). You say accelerating investment is causing sentiment to be bad. That doesn't really make any sense to me. Surely the opposite is actually the case? Businesses accelerate investment when they perceive a justifiable ROI (that means return on investment). By definition they need to have a positive sentiment - and anticipation - in order to justify making the investment. Unless those businesses are investing to comply with legislation I suppose. Investing in AI isn't a legislative requirement though.
c. You speak of a "survival rate". Where is the evidence that businesses like those held by HGT hold are not surviving? Which of its holdings or their competitors have shut their doors due to AI destroying them? I know there's a lot of waffle and talk but where are the actual liquidations? You warn of the wolf - but where are the dead sheep? Where are the declining sales? Declining profits? Where are the vibe coders undercutting the market introducing new software that up-ends the incumbents? Where are the customers cancelling their software subscriptions to the software that HGT sells because AI has designed a replacement in the blink of any eye? Where? Nowhere that's where. I maintain it is a nonsense - show me I'm wrong.
The way I see it even if you share some examples of companies in hard times (you quote Salesforce and yes their share price is down - but their GM % is 77.7% in the latest results and 73% historically - so not a good example of declining margins is it?).
Even if you do give some examples does it reflect on software as a whole? Software companies are in intense competition with each other and have been for decades. A continual arms race continues to improve the productivity of humans, where roadmaps and new features compete for attention and share of mind. AI is largely a marketing invention rebranding of office automation, database search capability and integration of systems and the provision of a new interface aka "the agent". Could AI actually change the world one day? Yes, possibly. But much of what's out there today is just a marketing rebrand - nothing more. Dressed up with fancy names of "Anthropic" and "Claude" but a Sheep in Wolves Clothing.
I think it is fair to say that the uncertainty around terminal value for software businesses is much higher now than it was at any time over the last 20 years due to the advancement of AI and the uncertainty of how quickly and in what direction AI will develop in the future and how that impacts software companies’ terminal value in years to come.
Software companies have always invested in R&D, but this will have to accelerate significantly in the future to try to defend their moat while at the same time they might not be able to pass that cost on to customers, given the competition with AI native solutions. This could lead to margin erosion for software companies.
It is not that software companies have gone out of business already due to AI, but that this might happen in the future as AI gets better and we might even get to agentic AI, which could replace whole software subsectors. Nobody knows if that will happen in a year or never.
There is therefore a lot on uncertainty around how software will develop over the next 5 years that could have a detrimental or negligible impact for software valuations. I’m not in the game of taking a flyer on how this will play out when even people at the coal face of AI and software don’t have a strong view. If HG and the other software investors were so convinced that certain software businesses were immune to AI then why haven’t we seen an uptick in take privates of software companies in the LTM, given public valuations have come down and are well below private valuations as evidenced by HG’s own valuations of its portfolio companies.
I think the point is that sell-off has been indiscriminate and does not differentiate between the winners and losers. Generic software (gaming is a good example) is clearly going to suffer, but HGT clearly believe their software is in the winner camp - highly specialised, mission critical software which is not going to be replaced by AI.
Indeed on the AI point, you'd actually expect AI to improve margins if it means they can develop their software more cheaply and add additional features which they can charge for.
Agree that 24x multiple still not particularly cheap, but with HGT on a 30% you're effectively getting these companies on a 16x multiple, which seems cheap for those growing, recurring revenues.
I heard a commentator speak about how gaming software has been deeply affected by AI. Is there any evidence to that being true? It's not an area I have looked at but their (and your) comment did pique my interest.
I'm kind of aware of games like "Call of Duty" having a massive following and brand value don't they? People spend hours and hours playing on them with their pals and they spend credits on in-game "power ups" and stuff. A lot of it is subscription too - you don't just buy a game - like I used to. Then you also buy merchandise and visit theme parks. Seems unlikely to me that vibe coders are building an alternative to "Call of Duty" in the blink of an eye and disrupting gaming forever?
To evidence that I just looked up Electronic Arts EA who I think produce call of duty are showing a record GM % of 79% in their latest results 31/3/26 and an op margin of 25.9% which is a much higher margin than 2021-2025. How is EA's market cap $50bn today up from $30bn in 2025 if gaming is being "replaced by AI"?
Be interested in any evidence you have to back up your "good example" or did you just hear the same commentator and assumed gaming software is suffering?
I don't have any evidence they are already suffering, but heard a number of tech analysts mentioning it a was an area at risk. I mentioned it just as an example of generic software that could be at risk and don't have particularly strong views either way, but logically it seems that generic games will be a lot easier and cheaper to make going forward. Where there is a strong franchise such as your Call of Duty exmaple, then that's probably a different story
Gaming software is a bit of a “hits” business. EA has just had a “miss” as a result of its Battlefield game, but there is no suggestion that the Saudi PIF take private will be adjusted. I’d say that gaming is a particular software sub sector to which general b2b trends might not apply.
Only cheap compared to historical multiples and assuming (a) all portfolio companies survive the AI disruption (b) Terminal Value remains unencumbered and (c) multiple compression doesn’t have further to run despite HG using 1/3 public comps and 2/3 private comps to value its portfolio companies. These are very big assumptions! Salesforce is trading at 11x EBITDA. Just saying…
The current share price of HGT at circa 30% discount to NAV which implies an underlying portfolio valuation of circa 19 times EBITDA still looks expensive to me. There is definitely a lot more uncertainty about the business models, moats and terminals value of SaaS companies given what Claude can do (and who knows what they or other models will be able to do in 5 years time) and the current valuation of the HGT portfolio in the mid 20s EV / EBITDA looks very high and a 30% discount to NAV to take the multiple down to 19x doesn’t materially change that (when calculating the 19x multiple implied by the share price you need to apply the 30% discount to the equity valuation of the portfolio (i.e. the part in excess of the 7x EBITDA of leverage at underlying portfolio level)). Closed end funds typically trade at a 10-20% discount to NAV whereas the current share price of HGT still implies in my view a material premium to the real underlying value.
I think the point is that if HGT indeed holds the type of software that is not at risk from being replaced by AI (deeply embedded, domain expertise, mission critical) then these type of companies should be trading at high mutiples, certainly well over 20. The market will understandably pay up for the combination of recurring revenues, growth, high margins and operational gearing. It seems to me that the market hasn't yet filtered between the winners and losers, with everything marked down, which creates something of an opportunity in my view. I would probably put Salesforce down as more generic software since it basically serves all industries. The more specialist software will typically service a very specific industry or sub sector - without detailed knowledge of the sector it will be difficult to break into these areas - with or without AI.
Salesforce is in trouble because their product is easy to copy, the data (for the most part) doesn't need to be exact, they haven't invested in it and they don't own the data. I had the misfortune of using it this week, not having looked at it for a few years and it has barely changed/improved. (And it wasn't good previously!) Companies who own the data and/or the data needs to be precise - think RELX, Sage, say, are still in a strong place in my view.
I went to the Salesforce "dreamforce" event at the London Excel last year. I was awed by the 40,000 attendees there. Their presentation and collaboration with Aston Martin was exemplary, absolute marketing gold dust. Those 40,000 people were mainly users and there was an incredible level of commitment to their product from what I could see.
However, like you, the quality of execution of Salesforce leaves a lot to be desired and as a user it feels incredibly clunky. Misfortune isn't an inaccurate description of having to use it.
Who do you think has copied Salesforce more successfully? Rivals like Dynamics CRM and Hubspot appear to be just as bad and I can't think of another that has the level of traction and breadth of functionality.
If clunky interfaces is the achilles heel of Salesforce could Agentic AI (Einstein I think they call it) not be its saviour?
I have been quite technical in previous lives (including working at a precursor to Salesforce). For the most part, Salesforce is just a reasonably good data model with a clunky UI on top. The data model is very easy to replicate - there's no IP there - and let's be honest, you wouldn't want to replicate the UI. The vulnerability in my mind is not from competing software but from DIY. I am very confident I could vibe code the majority of it and find someone better than me to do the hard bits.
I think your underestimating the stickyness of the customers once they're in the eco system, API integrations into other 3rd party software, 5 year contract terms. Migrating to something else is a big event and if companies can avoid it they will. Big system change in companies of all size is problematic and expensive.
I think software is oversold. Yes some companies will fall away but many will be left standing and reaping continuing annual reveunes until sentiment turns again and valuations normalise.
Maybe. But those apis are now super easy to vibe code (and much easier to change now when the apis change). And those 5 year contracts don't last forever - they keep needing to be topped up.... I admit to some bias here as I have worked with them a number of times over the years and rarely had a good experience.
I'm yet to see how many internal employees have time (or inclination) to vibe code a replacement. Not seeing it yet! It may come...latest earnings of software companies would indicate its business as usual. This can always change of course but could end up like waiting for the AI play to explode. Not happening yet.
Given that analysis, buying MSFT at 15x2026 EBITDA seems to me to be an absolute steal - usually traded at low 20s EBITDA multiple over the last few years. One of the most heavily moated companies around. You can worry about AI capex etc. But the AI hit has been taken in the sp and has come off the bottom. Been a 25% pa compounder on TSR over last 10 years. For me, buying US tech compounders when depressed has been a great “buy, hold and admire” play not requiring much effort.
Although market multiples have been impacted by AI replication concerns, that seems not to have fed through to the M&A market (or at least yet). SOftware Equity Group publishes aggregated software market data for M&A (probably mainly the US market) and reports an average 6.3x EV/TTM revs for q1 26, down from an exceptional 6.9x for q4 2025, but up from 6.1x in q1 2025 - so not a huge variation. Of course, companies impacted by AI might delay M&A if the owners can’t get the number they want ie it takes a while for the M&A market to adjust whereas the stock market is far quicker.
AI reports ranges of multiples for rule of 30 and rule of 40 companies (based on the SEG database I think). For rule of 40, the expected spread of EBITDA multiples is 25-35x, which implies, as one shd expect in PE fund NAVs, a decent uplift when they exit stuff. So it seems to me that HGT is in decent shape. I hold HGT as a long term play. The valuation discount might take a while to reverse. I notice some TR1s from savvy investors who perceive value here.
OB - very interesting podcast on PCT trust. Few words on software and very interesting on AI in general.
https://youtu.be/8_XbnRrV4xw?si=LpmdH_rXuYAA0H87
the following comment in the article is not correct: ‘Each £1 of HGT shares is earning nearly 30p per year’. The HGT portfolio valuation is £3.2bn which at a 24x EBITDA valuation implies an EBITDA of around £130 million per year. The market cap of HGT is £1.7bn, therefore this implies an EBITDA yield of around 8% and an earnings yield of likely between 4 and 6% per annum (and not 30%).
the following comment about the following comment in the article is not correct: Each £1 of HGT shares is earning nearly 30p per year’. The HGT portfolio valuation is £3.2bn which at a 24x EBITDA valuation implies an EBITDA of around £130 million per year. The market cap of HGT is £1.7bn, therefore this implies an EBITDA yield of around 8% and an earnings yield of likely between 4 and 6% per annum (and not 30%).
24X is how many times its earnings people are prepared to pay to buy a business (aka the exit price).
34% EBITDA margin is how much, on average, each of the holdings earn per year.
One’s a profit figure, the other is a purchase figure.
Hope that helps clarify.
OB
But the statement ‘each £1 of HGT shares is earning nearly 30p per year’ is wrong. The EBITDA Margin of the underlying business is indeed 34%, but the HGT portfolio generates around £130m of EBITDA per year (£3.2bn portfolio valuation valued at 24x means circa £130m EBiTDA). The market cap of HGT is £1.7bn which means that the portfolio EBiTDA as a percentage of market cap is circa 8%. Hence every £1 of HGT shares earns circa 8p of EBITDA per year and probably between 4 and 6p in net income.
The sentence was referring to EBITDA which means before interest, tax, depreciation and amortisation and I used the word "EBITDA" numerous times before saying "each £1 of HGT shares is earning nearly 30p per year".
But I take your point that it could be misconstrued so I've revised the sentence to say "each £1 of HGT shares is earning nearly 30p of EBITDA per year" to provide better clarity.
OB
It is not correct, £1 of HGT shares is not earning 30p in EBITDA each year. The £156m in trading result in the valuation bridge table is not the quarterly or semi-annual EBITDA produced by the portfolio, it is the increase in portfolio valuation as a result of increased portfolio EBITDA times the 24x EV/EBITDA valuation multiple which means an approx £6-7m increase in portfolio EBITDA.
The HGT portfolio is valued at £3.2bn which at an extremely high 24x valuation multiple means that it is producing circa £130m in EBITDA per year. When you divide £130m by the HGT market cap of £1.7bn, you get a circa 8% EBITDA yield which means that £1 of HGT shares produces 8p in EBITDA per year.
I assure you it is correct. You are still confusing earnings with valuation.
Earnings is what you earn. Valuation is what do or could sell it for.
Trading EBITDA is £154m for 1Q26.
£154m x 4 = £616m on a portfolio of £3077m, where 2400/3077 x 616 = £480m is attributable back to the £2400m NAV of HGT.
That's 20p in the pound isn't it? 480/2400 right?
But since you are only paying £1.62bn to earn £0.48bn EBITDA that’s 29.62% (480/1620)
The discount to NAV means HGT is generating a trading EBITDA of NEARLY 30p in the pound each year!
OB
Trading EBITDA is not £154m per quarter. The £154m is the increase in portfolio valuation as a result of trading. To get to the £154m increase in valuation you need an increase in portfolio EBITDA by £6-7m which at 24x EV / EBITDA valuation results in £154m valuation increase in portfolio valuation.
The annual EBITDA of the HGT portfolio is between £130 and £140m per year which is approx £33-35m per quarter.
Please don’t make any investment decisions based on the assumption that every £1 of HGT share price earns 30p of annual EBITDA as the reality is that it is (in my estimation) around 8p per £1 of HGT shares.
Also,,by referencing it to earning per £1 of HGT shares you bring valuation into the equation.
This. The OB is mixing up the way they report earnings growth with earnings.
Also: they report EBITDA growth INCLUDING M&A; it's not organic growth, so careful with Rule of 40.
Separately (and missed from the analysis above): their average business had a crazy 7.1x net debt/EBITDA at the end of 2021. I'm REALLY surprised more didn't go bust/equity valuations didn't crater when rates rose
No mix up. I was referring to EBITDA earnings but used the word "earnings" loosely and have added the word "EBITDA" for better clarity.
Thanks for raising the point about the level of debt from 5 years ago which you describe as "crazy". In that year HGT raised $125m junior bond financing to support its portfolio. That ratio went even "crazier" in 2022 hitting 8X and today it's above 2021 levels at 7.2X. I'm struggling to understand the point you're trying to make about the relevance of the 2021 ratio of net debt to EBITDA.
OB
I don’t know why you want to get involved here. Sentiment around software is so bad due to (a) uncertainty around terminal value, which will not be resolved in the short term (b) requirement to accelerate investment in AI capabilities that will reduce margins in the near term (c) current software multiples (24x for HGT) still do not reflect a particular pessimistic survival rate suggesting room for further downside. Just my 2 cents
Teubsch, did you even read the article? I don't know why you don't know if you did read it.
To your points
a. "uncertainty around terminal value, which will not be resolved in the short term" - When was there ever certainty? The sentence is meaningless as you're basically saying crystal balls are not about to be invented and the future is uncertain. Certainly is!
b. Requirement to accelerate investment. Yes, HGT have accelerated AI investment and their margins have grown (read the article). You say accelerating investment is causing sentiment to be bad. That doesn't really make any sense to me. Surely the opposite is actually the case? Businesses accelerate investment when they perceive a justifiable ROI (that means return on investment). By definition they need to have a positive sentiment - and anticipation - in order to justify making the investment. Unless those businesses are investing to comply with legislation I suppose. Investing in AI isn't a legislative requirement though.
c. You speak of a "survival rate". Where is the evidence that businesses like those held by HGT hold are not surviving? Which of its holdings or their competitors have shut their doors due to AI destroying them? I know there's a lot of waffle and talk but where are the actual liquidations? You warn of the wolf - but where are the dead sheep? Where are the declining sales? Declining profits? Where are the vibe coders undercutting the market introducing new software that up-ends the incumbents? Where are the customers cancelling their software subscriptions to the software that HGT sells because AI has designed a replacement in the blink of any eye? Where? Nowhere that's where. I maintain it is a nonsense - show me I'm wrong.
The way I see it even if you share some examples of companies in hard times (you quote Salesforce and yes their share price is down - but their GM % is 77.7% in the latest results and 73% historically - so not a good example of declining margins is it?).
Even if you do give some examples does it reflect on software as a whole? Software companies are in intense competition with each other and have been for decades. A continual arms race continues to improve the productivity of humans, where roadmaps and new features compete for attention and share of mind. AI is largely a marketing invention rebranding of office automation, database search capability and integration of systems and the provision of a new interface aka "the agent". Could AI actually change the world one day? Yes, possibly. But much of what's out there today is just a marketing rebrand - nothing more. Dressed up with fancy names of "Anthropic" and "Claude" but a Sheep in Wolves Clothing.
Plus Ca Change, Plus La Meme Chose.
OB
I think it is fair to say that the uncertainty around terminal value for software businesses is much higher now than it was at any time over the last 20 years due to the advancement of AI and the uncertainty of how quickly and in what direction AI will develop in the future and how that impacts software companies’ terminal value in years to come.
Software companies have always invested in R&D, but this will have to accelerate significantly in the future to try to defend their moat while at the same time they might not be able to pass that cost on to customers, given the competition with AI native solutions. This could lead to margin erosion for software companies.
It is not that software companies have gone out of business already due to AI, but that this might happen in the future as AI gets better and we might even get to agentic AI, which could replace whole software subsectors. Nobody knows if that will happen in a year or never.
There is therefore a lot on uncertainty around how software will develop over the next 5 years that could have a detrimental or negligible impact for software valuations. I’m not in the game of taking a flyer on how this will play out when even people at the coal face of AI and software don’t have a strong view. If HG and the other software investors were so convinced that certain software businesses were immune to AI then why haven’t we seen an uptick in take privates of software companies in the LTM, given public valuations have come down and are well below private valuations as evidenced by HG’s own valuations of its portfolio companies.
I think the point is that sell-off has been indiscriminate and does not differentiate between the winners and losers. Generic software (gaming is a good example) is clearly going to suffer, but HGT clearly believe their software is in the winner camp - highly specialised, mission critical software which is not going to be replaced by AI.
Indeed on the AI point, you'd actually expect AI to improve margins if it means they can develop their software more cheaply and add additional features which they can charge for.
Agree that 24x multiple still not particularly cheap, but with HGT on a 30% you're effectively getting these companies on a 16x multiple, which seems cheap for those growing, recurring revenues.
Hi Philip
I heard a commentator speak about how gaming software has been deeply affected by AI. Is there any evidence to that being true? It's not an area I have looked at but their (and your) comment did pique my interest.
I'm kind of aware of games like "Call of Duty" having a massive following and brand value don't they? People spend hours and hours playing on them with their pals and they spend credits on in-game "power ups" and stuff. A lot of it is subscription too - you don't just buy a game - like I used to. Then you also buy merchandise and visit theme parks. Seems unlikely to me that vibe coders are building an alternative to "Call of Duty" in the blink of an eye and disrupting gaming forever?
To evidence that I just looked up Electronic Arts EA who I think produce call of duty are showing a record GM % of 79% in their latest results 31/3/26 and an op margin of 25.9% which is a much higher margin than 2021-2025. How is EA's market cap $50bn today up from $30bn in 2025 if gaming is being "replaced by AI"?
Be interested in any evidence you have to back up your "good example" or did you just hear the same commentator and assumed gaming software is suffering?
OB
I don't have any evidence they are already suffering, but heard a number of tech analysts mentioning it a was an area at risk. I mentioned it just as an example of generic software that could be at risk and don't have particularly strong views either way, but logically it seems that generic games will be a lot easier and cheaper to make going forward. Where there is a strong franchise such as your Call of Duty exmaple, then that's probably a different story
Gaming software is a bit of a “hits” business. EA has just had a “miss” as a result of its Battlefield game, but there is no suggestion that the Saudi PIF take private will be adjusted. I’d say that gaming is a particular software sub sector to which general b2b trends might not apply.
Only cheap compared to historical multiples and assuming (a) all portfolio companies survive the AI disruption (b) Terminal Value remains unencumbered and (c) multiple compression doesn’t have further to run despite HG using 1/3 public comps and 2/3 private comps to value its portfolio companies. These are very big assumptions! Salesforce is trading at 11x EBITDA. Just saying…
The current share price of HGT at circa 30% discount to NAV which implies an underlying portfolio valuation of circa 19 times EBITDA still looks expensive to me. There is definitely a lot more uncertainty about the business models, moats and terminals value of SaaS companies given what Claude can do (and who knows what they or other models will be able to do in 5 years time) and the current valuation of the HGT portfolio in the mid 20s EV / EBITDA looks very high and a 30% discount to NAV to take the multiple down to 19x doesn’t materially change that (when calculating the 19x multiple implied by the share price you need to apply the 30% discount to the equity valuation of the portfolio (i.e. the part in excess of the 7x EBITDA of leverage at underlying portfolio level)). Closed end funds typically trade at a 10-20% discount to NAV whereas the current share price of HGT still implies in my view a material premium to the real underlying value.
I think the point is that if HGT indeed holds the type of software that is not at risk from being replaced by AI (deeply embedded, domain expertise, mission critical) then these type of companies should be trading at high mutiples, certainly well over 20. The market will understandably pay up for the combination of recurring revenues, growth, high margins and operational gearing. It seems to me that the market hasn't yet filtered between the winners and losers, with everything marked down, which creates something of an opportunity in my view. I would probably put Salesforce down as more generic software since it basically serves all industries. The more specialist software will typically service a very specific industry or sub sector - without detailed knowledge of the sector it will be difficult to break into these areas - with or without AI.
Salesforce is in trouble because their product is easy to copy, the data (for the most part) doesn't need to be exact, they haven't invested in it and they don't own the data. I had the misfortune of using it this week, not having looked at it for a few years and it has barely changed/improved. (And it wasn't good previously!) Companies who own the data and/or the data needs to be precise - think RELX, Sage, say, are still in a strong place in my view.
I went to the Salesforce "dreamforce" event at the London Excel last year. I was awed by the 40,000 attendees there. Their presentation and collaboration with Aston Martin was exemplary, absolute marketing gold dust. Those 40,000 people were mainly users and there was an incredible level of commitment to their product from what I could see.
However, like you, the quality of execution of Salesforce leaves a lot to be desired and as a user it feels incredibly clunky. Misfortune isn't an inaccurate description of having to use it.
Who do you think has copied Salesforce more successfully? Rivals like Dynamics CRM and Hubspot appear to be just as bad and I can't think of another that has the level of traction and breadth of functionality.
If clunky interfaces is the achilles heel of Salesforce could Agentic AI (Einstein I think they call it) not be its saviour?
OB
I have been quite technical in previous lives (including working at a precursor to Salesforce). For the most part, Salesforce is just a reasonably good data model with a clunky UI on top. The data model is very easy to replicate - there's no IP there - and let's be honest, you wouldn't want to replicate the UI. The vulnerability in my mind is not from competing software but from DIY. I am very confident I could vibe code the majority of it and find someone better than me to do the hard bits.
I think your underestimating the stickyness of the customers once they're in the eco system, API integrations into other 3rd party software, 5 year contract terms. Migrating to something else is a big event and if companies can avoid it they will. Big system change in companies of all size is problematic and expensive.
I think software is oversold. Yes some companies will fall away but many will be left standing and reaping continuing annual reveunes until sentiment turns again and valuations normalise.
Maybe. But those apis are now super easy to vibe code (and much easier to change now when the apis change). And those 5 year contracts don't last forever - they keep needing to be topped up.... I admit to some bias here as I have worked with them a number of times over the years and rarely had a good experience.
I'm yet to see how many internal employees have time (or inclination) to vibe code a replacement. Not seeing it yet! It may come...latest earnings of software companies would indicate its business as usual. This can always change of course but could end up like waiting for the AI play to explode. Not happening yet.
I'm not really invested software but am tempted.
Given that analysis, buying MSFT at 15x2026 EBITDA seems to me to be an absolute steal - usually traded at low 20s EBITDA multiple over the last few years. One of the most heavily moated companies around. You can worry about AI capex etc. But the AI hit has been taken in the sp and has come off the bottom. Been a 25% pa compounder on TSR over last 10 years. For me, buying US tech compounders when depressed has been a great “buy, hold and admire” play not requiring much effort.
Although market multiples have been impacted by AI replication concerns, that seems not to have fed through to the M&A market (or at least yet). SOftware Equity Group publishes aggregated software market data for M&A (probably mainly the US market) and reports an average 6.3x EV/TTM revs for q1 26, down from an exceptional 6.9x for q4 2025, but up from 6.1x in q1 2025 - so not a huge variation. Of course, companies impacted by AI might delay M&A if the owners can’t get the number they want ie it takes a while for the M&A market to adjust whereas the stock market is far quicker.
AI reports ranges of multiples for rule of 30 and rule of 40 companies (based on the SEG database I think). For rule of 40, the expected spread of EBITDA multiples is 25-35x, which implies, as one shd expect in PE fund NAVs, a decent uplift when they exit stuff. So it seems to me that HGT is in decent shape. I hold HGT as a long term play. The valuation discount might take a while to reverse. I notice some TR1s from savvy investors who perceive value here.
Oh Claude 😱 🧙
https://www.theguardian.com/technology/2026/apr/29/claude-ai-deletes-firm-database
Citywire artcile and vid on HGT
https://citywire.com/investment-trust-insider/news/hgcapital-s-finch-ai-will-sift-software-winners-from-losers/a2489026