• Samsung reported good results as smartphones and legacy DRAM fared better than expected, but the key catalyst for recovery which is the qualification of Samsung’s HBM4 at Nvidia, remains a hurdle yet to be cleared.
• Preliminary Q1 25 revenues and operating profits came in at KRW79tn / KRW6.6tn ahead of estimates of KRW76.6tn / KRW5.7tn.
• This was largely unexpected as demand for its new flagship smartphone seems to have been somewhat better than forecast while Chinese customers have been stockpiling commodity DRAM due to the geopolitical uncertainty.
• This pertains to both the potential for further restrictions on technology imports from the US Department of Commerce and the uncertainty around the new tariff regimes.
• This serves as a solid foundation for the recovery but the elephant in the room remains unaddressed.
• This is Samsung’s ability to compete in high bandwidth memory (HBM) that is needed in AI data centres to support both training and inference.
• Memory is often the bottleneck in the data centre rather than GPU speed or capacity, and so meeting or exceeding the required criteria is crucial to be a supplier.
• This is where Samsung has gone badly wrong as its HBM3e product has not been good enough to qualify with Nvidia as a supplier.
• Samsung has effectively given up on 3^rd generation and is putting all of its efforts into HBM4 which will start to take over from HBM3e towards the end of 2025 with volume in 2026.
• This failure has cost Samsung 40% of its market capitalisation creating an opportunity as its history has shown that it has the depth of corporate character to recover from substantial setbacks.
• This is why I own the shares and I think that over the next 12 to 24 months it will qualify with Nvidia, take share in HBM4 and return to around KRW85,000.
• I own a position in Samsung Electronics and I am looking for qualification at Nvidia as the catalyst for a rally in the share price.

Microsoft Copilot – Memory marketing.

• Microsoft celebrated its 50^th birthday by launching the next version of Copilot which it hopes will take Copilot from an oddity that is present mostly on the new long battery life Arm Windows laptops to something that has greater use for both consumer and enterprise.

• • First, Consumer: where a slew of new features have been launched as well as the ability to customise the look and feel of the agent.
  • Copilot will now be able to see what is on the screen, interact with supported apps (like Photoshop) making them easier to use and help with planning trips, shopping and so on.
  • Microsoft is also emulating OpenAI (and everyone else) by launching a Deep Research function bringing it up to date with everyone else.
  • Microsoft also demonstrated its agent actually doing things like filling in online forms with more abilities promised in future updates.
  • These agents will now have (with user permission) the ability to become more customised to understand the user’s preferences.
  • Microsoft refers to this as memory which is a neat marketing trick as short-term memory is a known problem that AI agents struggle with.
  • Microsoft has not solved this problem with the new Copilot as all it is really doing is using the knowledge graph of the user’s profile to adjust the weights of the model that is being used.
  • Second, Enterprise: where the launches are aimed at developers who will be developing custom agents for companies.
  • This is Microsoft enhancing its play for the ecosystem and at the same time deepening its divorce from OpenAI.
  • Microsoft no longer believes it needs to be at the cutting edge of AI performance and appears to me to be embracing some of the realities that its competitors continue to deny.
  • The idea now is not to have one agent that does everything but many agents all of whom are trained to do one task but to a high standard.
  • There is no reason why these can not work together which is what the new multi-agent framework is about.
  • Using this, developers can put together specific use cases that use numerous agents that together can complete a more complex task.
  • This is very similar to Nvidia’s Nvidia Inference Microservices (NIMs) and its AI foundry with the main difference being that Microsoft’s will be agnostic to the silicon it runs on.
• These launches are indicative of Microsoft’s evolving philosophy towards AI where it has publicly said that AI models are commoditising and that it does not need to be at the bleeding edge.
• I think that this is a result of its rapidly souring relationship with OpenAI as well as the realisation that the kind of dependency on OpenAI it was espousing was creating significant risk, especially given the precarious nature of OpenAI’s governance (see here).
• Although the case for agents is yet to be proven, Microsoft is in a good position to leverage the high share of personal computing and enterprise software to be one of the major players in the AI ecosystem.
• There still remains everything to play for and I suspect that 2025 will be all about wooing developers into building on these platforms that will determine the winners and losers of the AI Ecosystem.

• Another set of good results drove Palantir shares to new highs, but this was not the sort of blow-out that Nvidia has been reporting and with the shares 3x as expensive and rampant dilution, one is not getting much for one’s money.
• Q3 24 revenues / GAAP EPS were $726m (up 30% YoY) / $0.06 ahead of estimates of $703m / $0.05 and went on to raise guidance for FY 2024.
• 2024 revenues are expected to be $2.805bn – $2.809bn a range of just 0.1% from top to bottom meaning that Palantir already knows exactly what Q4 24 will look like.
• The current consensus is 2024 revenue of $2.76bn making this a beat of 1.7% which in my opinion is not meaningful in any way.
• Palantir’s business remains very US-centric with US commercial growing by 54% YoY and US revenues by 44% overall.
• However, this means that the rest of the world is doing very little where revenues expanded by just 7% YoY to $226m in Q3 24.
• Foreign revenues have been a problem for a while and some time ago, Palantir decided to pull back its investments and focus most of its efforts on the US.
• This is clearly going extremely well but the US is just 1/5^th of the global economy and Palantir will need to come back to this at some stage if it wants to keep the engine of growth going.
• Given that its previous efforts failed to yield results, success the second time around is not a given.
• Q3 24 also saw more share dilution with the diluted share count now standing at 2460m shares in issue up 1.8% QoQ and 5.8% YoY which I continue to think is a huge problem for shareholders.
• This is because shareholders are being diluted faster than earnings estimates are going up meaning that an objective and numerical-based valuation of this company is still falling.
• For example, earlier this year after a great quarter where expectations were slightly raised, my valuation of the company fell from $15 per share to $12.
• There is every sign that dilution is going to continue as the company is just not raising its estimates fast enough to justify the increase in the share price that has been seen.
• The shares jumped by 14% in after-hours trading meaning that the company is now valued at $115.6bn.
• This leaves the company trading on more than 100x EV/EBIT which is absurd even for a company as great as this one is.
• Palantir’s position remains intact in that it has no real competition, and its products are so good that companies who decline to purchase the software do so for fear of becoming dependent on them.
• It was also one of the first to come up with a product for managing and running LLMs in a secure and controlled environment and so is catching part of the generative AI craze.
• The problem is that the revenue base is not nearly big enough to justify the valuation and if I was still holding the shares, I would sell them immediately.
• The case for Palantir shares at $6 (when I bought them) was fantastic which became indifference at $15 and dislike at $20 (when I sold them).
• What has happened since is pure narrative as the numbers have not moved that much in terms of performance relative to expectations, meaning that when the narrative runs out of steam, the shares could easily crater by 50% or more.
• I missed a double by selling early but remain perfectly happy to have moved on to better-valued territory.

Peloton FQ4 24 – The turn

• Peloton reported revenues that grew YoY dispelling fears of the company being in a death spiral which, combined with finally breaking even on a cash flow basis, gives hope that the ship has finally stopped sinking.
• The result was a 35% jump in the share price but given how far the shares have fallen from their 2021 high of almost $175, this is a tiny blip that can hardly be seen on the chart.
• Q4 revenues / EPS of $643.6m / LOSS$0.08 ahead of estimates of $628m / LOSS$0.17 and guided for FQ1 25 broadly in line with expectations.
• The real surprise came in both operating profit where the company almost broke even and in operating cash flow which posted a $32.7m cash inflow for FQ4 24.
• These are strong signs of stabilisation which combined with subscribers remaining loyal to the company, allows the focus to shift to the future.
• This is where the recruitment of a new CEO (yet to be announced) will be crucial.
• Now that the company is generating cash, the question will be what to do with that cash as there is a strong argument for running the company for cash and giving it back to the shareholders.
• The subscriber business has 3m subscribers that generate $1.6bn in revenues from which the company earns 70% gross margins which with a hardware business that breaks even and modest OPEX, it is possible to value the company at around $8bn with no growth at all.
• Take off $1bn for the debt and one ends up with a per share valuation of $19 compared to $4.55 (after rallying 35% on August 22^nd).
• This all depends on hardware no longer draining the business of value and a greatly reduced operating expenditure both of which are now well underway.
• It also relies on the new CEO not making bad strategic decisions and wasting the value that is being generated by the subscription business as the original management did on hardware.
• However, with the biggest risk factors receding into the rear-view mirror, the possibility of a major rerating is now in the cards.
• I have a half position in Peloton and am considering adding some more here as even with the 35% rally, it is still below where I originally bought it.

Snowflake FQ2 25 – Not enough

• Snowflake reported good results but the weak guidance and the bad PR from the cyberattack suffered earlier this year meant that the shares sold off underlining that this remains one to avoid.
• FQ2 25 revenues / EPS were $869m / $0.16 broadly in line with forecasts of $851m / $0.18 but guided weakly for the coming quarter.
• Here FQ3 25 revenues are expected to be $850m – $855m which is below $899m which is what has triggered the latest disappointment in a company where the story is all about growth.
• This is because even excluding stock-based compensation, the company is still barely profitable despite an annualised revenue run rate of over $3.2bn.
• Hence, to generate value for shareholders it needs to have a much larger revenue base over which to spread its huge operating expenditure.
• The company does generate cash and so there are no fears of its solvency but it is more a question of what this business is worth.
• Even after falling 14.7% on August 22^nd, the market capitalisation of the company remains $38.4bn which when adjusted for net cash gives a FY 25 EV / Sales of 11.0x.
• This would be fine if the company were generating excellent profits and cash flow but this is far from the case as the FY 2025 PER of 191.7x clearly indicates.
• Hence, even without the stigma of a cyberattack and the possibility of being disrupted by generative AI, I would not be enticed to own this one.

• Nvidia AI Foundry is a consequential announcement as it looks like it has been designed to maintain Nvidia’s dominance in AI when CUDA is no longer the control point in the AI ecosystem.
• However, I am pretty certain that the timing of the announcement and the complexity of the topic will ensure that no one pays very much attention to it.
• The best way to understand Nvidia AI Foundry is to compare it with TSMC (as Nvidia does in its materials).
• TSMC manufactures chips for fabless (no factory) semiconductor companies who design their chips, and send those designs to TSMC, which manufactures them on their behalf.
• Nvidia AI Foundry is directly analogous to TSMC in that it provides tools and services for AI developers to create their own models which are then packaged into NIMs (Nvidia Inference Microservices) which are optimised for the inference architecture that the developer wishes to run.
• A NIM is a customised model or models that sit inside a wrapper containing a standardised API meaning that 1)  multiple NIMs can be easily plugged together to create a specific service and, 2) they are optimised to run on Nvidia inference hardware.
• The NIMs are “manufactured” in a NIM Factory which is the equivalent of one of TSMC’s fabs and then the finished NIM is delivered to the developer to use in the service it wishes to offer to customers.
• The key to this is that while NIMs will run on non-Nvidia hardware, they will run so much faster and so much more efficiently on Nvidia hardware, that there is no point in using anything else.
• This means that if NIMs become the standard way to create and deploy AI services (generative and other), then Nvidia will have replaced the control point of CUDA with the control point of NIMs as nobody using a NIM is going to use non-Nvidia hardware.
• Hence, with NIMs as an industry standard, Nvidia will hold onto its 85% market share in training and inference as well as its 75%+ gross margins.
• Nvidia AI Factory is already off to a strong start with some heavyweight endorsement from Meta Platforms and Accenture with Meta Platforms timing the release of its new model LlaMa 3.1 to coincide with Nvidia AI Foundry Launch.
• Accenture has been tinkering with NIMs as an early partner but will now use the AI Foundry to create NIMs on behalf of its clients.
• Given the size of Accenture, this is a big endorsement and the CEO of Accenture, Julie Sweet was quoted directly in the announcement.
• Nvidia AI Foundry is launching with Meta’s Llama 3.1 update which is available now in the sizes of 8bn, 70bn and for the first time, 405bn.
• What is new here is that this is the first time that Meta has made its flagship model available to open source and, if Meta is to be believed, it ranks first in 7 out of 15 evaluations when compared to Nvidia’s Nemotron 4, GPT-4, GPT-4o and Claude 3.5 Sonnet.
• This is relevant because none of the other flagship models (other than Nemotron 4) are available to the open source which combined with being available on AI Foundry from day 1 should provide a draw for developers.
• Its initial use case in the AI Foundry is for synthetic data generation but I suspect that those that wish to compete with OpenAI or Anthropic directly will quickly create NIMs that do precisely that using Llama 3.1 405bn.
• This is really bad news for OpenAI, Anthropic, Google and so on as all of them are hoping to make money from something Meta has just given away for free.
• This could potentially accelerate the race to the bottom in terms of generative AI service pricing which is what I think will trigger the reset from AI hype to reality.
• This announcement also makes Nvidia’s strategy much clearer which is to remain the go-to place to develop AI services even when developers are no longer so focused on CUDA.
• This is what RFM Research refers to as AI Ecosystem 3.0 which is way ahead of where we are today (AI Ecosystem 1.0) where the control point is at the silicon development platform.
• Nvidia’s current dominance in this space provides it with an opportunity to migrate loyalty and preference for CUDA into loyalty and preference for the AI Foundry which is why I refer to it as CUDA II.
• Ironically, the announcement was made on the day that Nvidia’s shares tanked 8% in line with the rest of the technology sector which underlines that the level of attention and understanding of what AI Foundry implies is pretty much zero.
• Hence, if there is a big correction in AI-related stocks as reality bursts the hype bubble and AI Foundry really takes off, then there will be another opportunity to invest in Nvidia for those who missed it the first time around (like me).
• For now, I am sitting tight in my AI adjacencies of inference at the edge and nuclear power.

• Mike Lynch the ex-CEO of Autonomy has the last laugh in the 12-year battle over who was really to blame for HP’s calamitous acquisition of Autonomy which I have long believed was a result of HP’s incompetence and desperation as opposed to criminal activities by Mike Lynch and his team.
• In 2011, HP acquired UK software company Autonomy for $11.7bn and then 14 months later it wrote down the acquisition by $8.8bn (75%) claiming “serious accounting improprieties” and “outright misrepresentations”.
• HP then went on a 13-year legal crusade to blame the management of Autonomy for fraudulently inflating the value of the company claiming that HP’s management and its board were the honest victims of fraud.
• I have long viewed this as an effort to blame the staggering incompetence of HP’s board and its management on Autonomy because it was as clear as day that the company did not do its due diligence properly (see here to read my piece from 2012).
• The main reason for this view is that it in the two years that led up to the acquisition, I was working in investment banking where I covered Autonomy as a technology specialist in sales and trading which is the period that HP claims that the fraud took place.
• Autonomy’s detractors had been writing about Autonomy’s accounting for years and there had been the occasional sign that things were not quite right.
• The most common red flag was that cash flow in the quarterly reports often did not match profit which is quite unusual in a software company.
• The balance was often made up in subsequent quarters but critically, this and other issues caused people to wave the red flag.
• At the time it was clear to the market that Autonomy played fast and loose with the accounting rules, but this was more about smoothing the revenue over a year as opposed to fraudulently inflating the revenue numbers.
• The motivation for doing this would be to reduce the volatility in the share price as a result of the lumpier-than-normal fluctuations in revenue.
• To any reasonably prudent person, this would have made him doubly cautious when looking into the financial position of Autonomy.
• History has shown that a determined person can deceive a competent auditor without too much difficulty because it is simply not economical or timely to look under every single rock or pebble.
• In this case, there were more than enough red flags and so much money at stake to warrant a due diligence process that went beyond the audited figures.
• However, HP’s board and management were in such a rush to turn HP into a high-margin software company, that they decided to ignore the red flags and press ahead with the acquisition nonetheless and shareholders paid the price for it.
• To me, this said far more about HP than it did Autonomy and it is no surprise that a board and management clear-out followed in the subsequent months.
• To be fair to HP, it has had some success in prosecuting its claims in the civil court and Autonomy’s previous CFO has served 5 years after being found guilty of fraud, but I now wonder whether that conviction is safe.
• This is because Autonomy’s former VP Finance was also on trial with Lynch and was also found not guilty meaning that there is more to the defence other than Lynch being ignorant of the crime because he left these matters up to others as he testified.
• The net result is that this serves as a cautionary tale to any company panicking about being left behind or under pressure to make change with the message being that the due diligence process is there for a reason.

• Intel came out of its corner swinging taking on both Nvidia and Qualcomm in a keynote that was really all about how its latest processors stack up against the new upstarts.
• Intel was quick to refute Nvidia’s claim that Moore’s law is dead and if any of its data is accurate, there is some substance to this claim.
• This was on display with its latest data centre CPU Xeon 6 where eBay and SAP are supporting Intel’s claims.
• eBay claims that the Xeon gives a 25% improvement in performance per watt while SAP was able to squeeze out a 60% saving over the previous generation of Xeon.
• Intel also took a shot at the Nvidia H100 claiming that Gaudi 3 is 40% faster than Nvidia’s top-selling H100 processor and can deliver 2.3x the performance per dollar spent.
• The problem here is that the Hopper is no longer the benchmark as Nvidia launched Blackwell a couple of months ago against which Guadi 3 will be much less impressive.
• I suspect that Gaudi 3 will still be cheaper when it comes to performance per dollar when compared to Blackwell but this is also because everyone is so desperate to get their hands on Nvidia’s products that it can earn nearly 80% gross margins which is almost unheard of in semiconductors.
• Gaudi is a good option for the few that don’t want to use Nvidia or want to dual source, but it doesn’t have CUDA and as such, I don’t see it as a massive competitive threat.
• In AI PCs, Intel also came out swinging at Qualcomm which will be pretty gratifying for Qualcomm that Intel considers it to be such a threat.
• On the last two occasions that Qualcomm has taken a shot at this market, Intel didn’t even bother to acknowledge that the threat existed.
• It certainly seems to be the case that it is third time lucky for Windows on Arm.
• Here Intel claimed that the Lunar Lake CPU, GPU and NPU can all outperform the X Elite, which is credible, but crucially, Intel neglected to mention at what level of power consumption.
• Unlike its peers, Intel was brave enough to make power consumption comparisons against Meteor Lake, it said that Lunar Lake could achieve a 40% SoC power saving.
• Qualcomm’s test claims a 65% saving against Meteor Lake and so it’s a pretty safe bet that the X Elite will fare better on performance per watt.
• However, the question is how much and here there is no way to know until the devices get in the hands of consumers and testers and we see them work in real-world situations.
• Purely on the numbers if Meteor Lake is consuming 100 units of power, then Lunar Lake is claiming 60 and X Elite 35 implying that X Elite will still offer substantially longer battery life than Lunar Lake.
• It is clear that Intel is not going down without a fight, but Qualcomm is coming to market with Copilot+ PCs first and as such it is winning the PR battle against the x86 camp.
• Against Nvidia’s public presence juggernaut, Intel also faces an uphill battle meaning that this Computex has found other champions.
• Intel still has everything to do.

• Nvidia contemptuously swept aside both its critics and its competition and reported another huge set of results that confirmed that the AI frenzy and the bubble continue to grow and inflate.
• FQ1 25 revenues / Adj-EPS were $26.0bn / $6.12 ahead of forecasts of $24.6bn / $5.60 and Nvidia signalled that the strength will continue.
• FQ2 25 revenues are expected to be $28.0bn +/- 2% again ahead of consensus of $26.6bn.
• The real story of these results was FQ1 25 gross margin which came in at 78.9% (almost unheard of in the semiconductor industry) up 12.5% points from FQ1 24.
• This is the clearest signal that Nvidia can charge whatever it wants for its silicon and still, the customers will beat down the door and buy everything that it can make.
• Nvidia was also untroubled by one of its largest customers delaying its orders to wait for newer chips as there are more than enough other customers waiting to take AWS’s place in the queue.
• This supports my view that fiscal 2025 revenues will be determined by how much capacity Nvidia has reserved at TSMC as it is clear that it can currently sell everything that it makes at superb gross margins.
• It is also a signal that nobody wants to buy chips from the competition many of whom are struggling to make any sales unless they address a very specific niche in the market.
• The main driver once again was the data centre which grew by 26% QoQ and a massive 427% YoY as it is here where most of the investments are being made.
• Gaming, enterprise metaverse, automotive and robotics also grew well but are now a rounding error in the accounts with data centre making up 87% of total revenues.
• This means that the outlook for the short to medium term as well as the valuation of the shares is almost entirely dependent on the generative AI investments in the cloud and at the moment there is nothing but good news.
• However, at some point, it will become clear that the LLMs that power generative AI have no understanding of causality and remain unable to reason.
• This is important because these are two crucial abilities that the machines need to become truly intelligent.
• The hyperscalers and AI ecosystem proponents are leading the market to believe that LLMs have these characteristics even though the empirical evidence indicates the opposite.
• Hence at some point, there is going to be a reckoning when generative AI fails to live up to its promise and there is a correction just like there is every single time the market gets worked up in a frenzy like this.
• For better or for worse, Nvidia is on this rollercoaster and when the correction comes it will be unable to get off, but it will incur far less pain than anyone else.
• This is because Nvidia is the only one making real money out of generative AI at the moment and this will continue even in a downturn as generative AI has a lot of use cases even with the machines being unable to reason or understand causality.
• This is reflected in Nvidia’s valuation which is really not that bad as a result of the huge increases in both revenues and margins.
• With the shares at $1,000, Nvidia is on FY25 PER of 37x or so which is broadly in line with Microsoft, but Nvidia is growing much faster.
• This means that when the correction comes, Nvidia is likely to fall less than its peers making it the safest pure AI play in my opinion.
• However, I would continue to prefer to look more laterally on AI themes at things like nuclear power or enablers of inference at the edge like Qualcomm (and pretty soon MediaTek) as better ways of gaining exposure.
• I have positions in both and remain comfortable to sit tight.

• Although the real action was yesterday, Microsoft used its 2024 Build keynote to lay out its proposition to developers for using Microsoft tools to build AI and it did a far better job of it than Google and in half the time.
• Microsoft’s message was very clear in that it showed the AI technology stack from chip to service and then explained at each level how it provides the best tools and services.
• Its case is quite compelling compared to Google whose AI is just as good but dumped a series of product launches into its presentation and left the audience to figure out what they are and what to do with them.
• Microsoft had very little to announce that was really new but what it did do was advance tools towards general availability and announce new features that allow developers to build and customise more on its core products.
• For example, Microsoft is bringing in AMD to sit alongside Nvidia and its in-house accelerator as well as pretty much every foundation model that is available so that developers can pick and choose what they want to build on and which silicon they wish to execute on.
• Another example is GitHub Copilot which already has 1.8m users and 50,000 organisations engaged with it.
• Here the main news was the launch of extensions meaning that developers can customise the capabilities of their service based on Copilot with pretty much anything that they want.
• On Monday, Microsoft announced Team Copilot which functions like an executive assistant team member that tracks action items, takes minutes, manages a project timeline and so on.
• On Tuesday, Microsoft announced the tools to customise Team Copilot to one’s own specification using Microsoft modules or ones that one builds oneself.
• However, the one place that Microsoft always went back to was Azure AI Studio which is where Microsoft wants the developers to spend all of their time.
• Microsoft does not have a foundation model of its own as everything is based on Open AI’s GPT foundation and this is why, like Nvidia, it is making its play for the AI ecosystem further up the stack.
• Google, OpenAI, Anthropic, Meta and others have invested vast sums in creating their own foundation models and their play for the ecosystem is to lock developers into their models which are quite difficult to swap in and out.
• Microsoft’s position is different in that its proposition is that it does not care which foundation the developer uses as long as the developer uses Azure AI Studio and runs the training and potentially the inference on its infrastructure which is where it will make its money.
• This is how Microsoft intends to become one of the main tool providers for the development of AI and its experience as a tool provider was clearly evident.
• It did a good job of identifying where the use cases in the enterprise are going to be and has moved quickly to offer tools and services that will allow developers or companies themselves to address these use cases.
• These are in line with RFM Research’s conclusions.
• Microsoft might not be a leader in the nuts and bolts of AI or in driving the industry towards general intelligence but it is a leader in providing an easy-to-understand proposition that allows developers to create the AI that they want.
• In the short to medium term, this is going to be more valuable than spending billions on a moonshot and so this cements Microsoft’s position as the leader in the provision of enterprise software and services as AI increases its usage and penetration.
• That being said, Microsoft is very far from being the bargain that it once was and so this is not how I would invest in the AI trend.
• Instead, I would look more laterally like nuclear power or enablers of inference at the edge like Qualcomm (and pretty soon MediaTek) as better ways of gaining exposure.
• I have positions in both and remain comfortable to sit tight.

Alphabet: AI everywhere

• Alphabet reported strong results and spent its conference call reminding everyone that it is the best AI company even if it does a poor job of getting that message out to the media and investors.
• Q1 24 revenues-exTAC / EPS were $67.6bn / $1.89 ahead of estimates of $66.1bn / $1.53.
• Alphabet also broke precedent by announcing a $0.20 quarterly dividend (measly 0.4% annualised yield) as well as a $70bn share buyback which was well received.
• However, these are a sure sign that slower growth is ahead in the long term as companies tend to return money to investors when they can’t think of other places to invest it or want to support the shares.
• Search and YouTube once again performed well but the biggest jump was seen at Google Cloud where revenues grew by 28%.
• This sounds great until one looks at Microsoft (see below) where revenues grew even faster, undermining Google’s narrative that the world is coming to Google for AI.
• The reality here is that the current hype around what generative AI is capable of has lit a fire under everyone who offers AI compute for rent and I suspect that AWS will also see an acceleration when Amazon reports its Q1 24 results on 30^th April.
• In short, these results go someway towards helping Google dispel the narrative that it is a laggard in AI (which I have never believed) which combined with the good results is why the shares reacted so strongly.
• Alphabet is now enjoying a rerating meaning that the best of its performance is probably already behind us making me much more ambivalent to the company.
• However, there is space for it to rally some more for those who didn’t miss it (like I did) earlier this year.

Microsoft: Making the most of good times.

• Microsoft reported excellent results as demand for AI continues to grow but at the same time, capex will increase materially as Microsoft needs more data centres in which to train and run all of these large models.
• FQ3 24 revenues / EPS were $61.9bn / $2.94 ahead of estimates of $60.9bn / $2.83 which combined with a reacceleration of cloud growth underpinned a 5% increase in the shares after hours.
• This beat is not dissimilar to the one reported by Alphabet underlining that Alphabet additionally has the tailwind of a multiple re-rating and that Microsoft’s multiple is already pretty full.
• Star of the show was undoubtedly Azure where growth has picked up to 31% YoY which is quite an achievement given how large this business has become.
• Microsoft is claiming that it has taken market share implying that when all the reports are in, 31% YoY will prove to be the fastest growth across the big 3.
• The outlook for the rest of the year is also strong as commercial booking has also increased by 29% YoY meaning that strong growth looks like it will continue for a few quarters yet.
• This is fuelled by both AI and non-AI-related demand, but I am pretty sure that it is AI that is driving all of the outperformance.
• Microsoft has also done an excellent job at wrong footing Google in this latest AI craze and this has been reflected in the difference in valuation between the two companies although Google is doing a good job of closing that gap.
• I remain pretty ambivalent about Microsoft as the valuation already reflects the outlook but also has the added danger of taking a whack should the AI bubble burst which it surely will at some point.

Intel: Painful twenties.

• Reasonable results were overshadowed by weak guidance in a clear sign that the transition is far from complete and that Intel is continuing to miss out on the AI boom.
• Q1 24 revenues / EPS were $12.7bn / $0.18 where revenues were in line with estimates and profits ahead but Q2 24 guidance was weak which Intel blamed on a supply bottleneck.
• This bottleneck has occurred in the client business, and one would have expected its data centre business to pick up the slack given the strength coming from Google, Microsoft and undoubtedly Amazon.
• This has not happened and revenues / EPS for Q2 24 are expected to be around $13.0bn / $0.10 below estimates of $13.6bn / $0.24.
• Intel is optimistic for H2 2024, and it is pinning its hopes on Gaudi which is a new AI chip it has created to challenge Nvidia but like AMD and all of the other Nvidia wannabes, it doesn’t have the depth of development platform and tools available for developers.
• Hence, I am not convinced that Gaudi is going to see a lot of traction and would be cautious of believing estimates that Gaudi is going to pull Intel out of its current troubles.
• I also remain concerned about the impact that Qualcomm’s new X Elite line of processors is going to have on Intel’s market share as Qualcomm is pulling out all the stops to make sure that this third attempt to crack the laptop processor market actually works.
• By all available metrics, X Elite is much better than Intel when it comes to performance per watt of power consumed which is a critical metric in any battery-powered device.
• This is why I continue to struggle with Intel’s valuation which even after a 7.8% fall in after-hours trading is trading at around 15x 2025 PER.
• This is more expensive than Qualcomm leaving me in no doubt as to which one I should hold.
• I own Qualcomm and have no intention of selling it yet.

• Another set of excellent results confirms that for as long as the AI rollercoaster continues, Nvidia is the occupant in the train and will continue to experience a wild ride.
• FQ4 24 revenues / EPS were $22.1bn / $4.93 nicely ahead of forecasts of $20.4bn / $4.59 and there is the promise of more to come as there is no end in sight to the generative AI hype.
• Data centre was the star of the show posting revenues of $18.4bn up 27% QoQ and a massive 409% YoY as demand continues to far exceed supply.
• There are signs that the market is beginning to shift from training to inference as Nvidia estimates that 40% of its revenues in the last year are for inference rather than training.
• Furthermore, over half of Q4 24 revenue was from large cloud providers again underlining just how dominant they are, and how spectacularly they have failed to come up with in-house alternatives to Nvidia.
• This is on full display in Nvidia’s profitability numbers where gross margins rose to 76% (GAAP) and GAAP EPS was up 765% YoY which goes a long way to supporting Nvidia’s valuation.
• It also highlights that the ever-increasing number of competitors both in the merchant market and in-house remain unable to make a dent in Nvidia’s proposition.
• This is a result of its CUDA development platform which is far superior to anyone else’s and has become an industry standard and its breakneck product cadence which ensures it always has the most advanced products in the market.
• The good news was not just limited to data centre as both automotive and gaming beat expectations demonstrating a company that is firing on all cylinders.
• Outlook was also good with FQ1 25 revenues expected to be $23.5bn – $24.5bn ($24bn) with steady gross margins ahead of forecasts at $22.2bn.
• The reality here is that Nvidia has more demand than it can deal with, and so its revenues for the next few quarters will be determined by how much capacity it has purchased from TSMC rather than how much customers have ordered.
• Hence, visibility should be pretty good which is reflected in the pretty tight 4% range it has given for its revenue guidance.
• However, I suspect that it has understated its expectations slightly because it is always better to do better than expected rather than promise the moon and miss.
• Consequently, I expect that the next few quarters will be characterised by a beat and raise but by decreasing amounts as the momentum becomes harder to sustain as a result of the large numbers that are now in play.
• These results were characterised by two main themes:
• First, Nvidia’s contention that accelerated computing and generative AI have hit a tipping point and second, that the mantle of growth is starting to be underpinned by inference rather than training.
• This theme was recently echoed in Jensen’s presentation at the World Government Summit in Dubai where he made the case for accelerated computing and how being able to rent GPUs on a demand basis can benefit the growing start-up ecosystem in the Middle East.
• At the moment all of the inference is being carried out in the cloud and almost all of it is also being executed on Nvidia silicon.
• RFM research has concluded that 2024 is going to be the year of inference and as almost all of it is not ready to move out from the data centre, Nvidia will continue to benefit.
• One thing is clear from the results and conference call which is that Nvidia is the picks and shovels of generative AI and, try as they might, no one else is currently getting a look in.
• This means that for better or for worse, Nvidia is stuck on this rollercoaster and there is no way for it to get off should the bubble burst.
• However, for the next few quarters, there is not a cloud in the sky and I expect that Nvidia will continue to perform well and the shares will reflect that in their valuation.
• Despite the stratospheric rise, the valuation is not unreasonable at 32.5x 2024 and 28.3x 2025 PER making Nvidia a far better proposition than many of the start-ups who are demanding billions in valuations but with no revenues let alone profits.
• It was the tools companies that made money out of the gold rush in the USA and the diamond rush in South Africa and for this rush, Nvidia is the only tool company in town.
• I remain very nervous with regard to the bubble popping as there is ample evidence that these models cannot do many of the things that people expect them to and so I am inclined to sit tight and wait for a crash.
• At that time, the babies will have been thrown out with the bathwater and there will be bargains on offer.