• 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 $40bn to spend on compute will ensure that OpenAI does not have to become efficient like its Chinese competitors meaning that when the money runs out, it is likely to be caught out resulting in it being forced to sell itself to a rich competitor or one of its backers.
• OpenAI is raising $40bn at a valuation of around $300bn from which $10bn ($7.5bn from SoftBank and $2.5bn from an investor syndicate) is going in now.
• The other $30bn will go in later this year with $22.5bn from SoftBank and $7.4bn from the investor syndicate.
• Almost all of this is likely to be spent on compute as OpenAI continues to make its models bigger, train its models harder and use more data in its quest to create artificial general intelligence (AGI).
• This is the point at which the machines become more intelligent than humans and take over more than 90% of all economically active tasks.
• The assumption being made by OpenAI’s investors is that it will be able to reach AGI before anyone else and do so in a relatively short period of time.
• This would give it exclusive access to arguably the most valuable asset ever created, and assuming that it monetises this asset, a fair valuation in the many trillions of dollars.
• However, there are two caveats:
  • First, No AGI: as RFM research and all of the available evidence suggests that a system that is based on statistical pattern recognition will never be truly intelligent.
  • AGI requires an understanding of causality and the ability to distinguish between relationships that are merely correlated by chance and those where one affects the other.
  • It is this shortcoming that prevents the machines from understanding the causal nature of the tasks that they have been asked to complete and is why they make things up, get things wrong and are generally unreliable.
  • As a result, RFM Research has concluded that the approach that OpenAI is taking will never produce AGI and when this realisation hits the market, the money will dry up and OpenAI will be forcibly acquired.
  • Second, Corporate Governance: OpenAI remains a hot mess of a for-profit entity being governed by a board that is tasked with overseeing the creation of AGI and its distribution to everyone for the benefit of all mankind.
  • There is a large conflict here which very nearly caused the company to completely implode and may well do so again.
  • Furthermore, the rapidly souring relationship between Microsoft and OpenAI means that if conflicts emerge, there is now a large and grumpy shareholder capable of making a lot of trouble.
• The net result is that my view on OpenAI is very binary.
• If one believes that OpenAI will create AGI and do so before anyone else, then SoftBank is making a truly great investment.
• However, if one believes, and as the evidence indicates, that statistical-based systems will never create true intelligence and that the market for models is commoditising, then the real value of OpenAI is far below $300bn and may even be $0.
• The real winner from this investment is CoreWeave whose difficult IPO has its first day of trading today.
• This is because it appears likely that OpenAI will spend a portion of the $40bn on CoreWeave compute services taking up the slack that Microsoft has created by reducing its deal with CoreWeave.
• This will be a badly needed shot in the arm as CoreWeave has had to downsize its offering as well as reduce the valuation from $32bn to $23bn.
• This puts the company on 15x EV / Revenues for 2024 as it has a net debt position of ($6.6bn) which is pretty expensive for a company that provides infrastructure for AI rather than the AI itself.
• However, sentiment has rapidly soured on CoreWeave meaning that if it trades badly on the opening, it could be worth having a look at.
• Alternatively, Nebius which is in a similar business and at an earlier stage of its roll-out (given 2024 revenues of $118m), but has no debt and $2.5bn in net cash may be a much better and safer option.
• This (plus Nvidia and inference and nuclear power) is where I would be much more interested in looking rather than in the hugely valued providers of foundation models which are showing every sign of turning into commodities.

GTC Update: Nvidia Dynamo – the most critical launch of 2025. 

• GTC 2025 is in full swing and as more details emerge about Dynamo, it is clear to me that this is by far Nvidia’s most important launch of 2025.
• Dynamo is an operating system for a data centre that is producing tokens (inference) for a generative AI service.
• Dynamo looks at the system of GPUs, memory and networking and works out the most efficient way to manage these resources based on the nature of the requests that are coming in.
• It allows the data centre to produce as many tokens as possible for its resources, which maximises revenue given that industry standard pricing is $ per million tokens.
• Dynamo makes more sense now because of the new family of “reasoning models” that improve performance by massively increasing the number of tokens that they produce per request.
• This means that inference is going to quickly become the largest function of the data centre as RFM has been predicting for some time.
• To be as dominant in inference, Nvidia needs more than CUDA as I think that CUDA is much less of a control point in inference than it is in training.
• Enter Dynamo which promises to do for inference what CUDA has done for training.
• If it is as good as Nvidia claims (see here) then Dynamo users are going to become more competitive given the efficiency improvement.
• However, Dynamo is only likely to work really well on Nvidia hardware and the fact that anyone would want to run it on competing silicon does not seem to have been considered.
• “In theory, you could do that as we have made it available to open source” was the response to the question, but obviously, this is not what it has been built for.
• Consequently, if Dynamo proves to be very popular with clients, it is likely to raise barriers for data centre owners who are considering running inference on competing silicon.
• Given that CUDA has been around for over 20 years and Dynamo is brand new, Dynamo has a long way to go if it is going to replicate CUDA’s effect in training, but the seeds have been sown.
• Competitors expecting to take a bite out of Nvidia when it comes to inference need to act quickly if they don’t want the slightly open door of inference to be quickly slammed in their faces.

Samsung: Bet on HBM4. 

• Samsung apologised yet again at its AGM for its lack-lustre performance in memory and has promised to do better in 2025 setting the shares up for a substantial re-rating if it is successful.
• Like the Note 7 that spontaneously exploded, Samsung’s memory problem is simple in that its memory offering for AI was not good enough allowing SK Hynix and Micron to humiliate it at its own game.
• This has greatly damaged both Samsung’s performance and its reputation as memory for AI is one of the hottest areas in semiconductors right now.
• If one takes Nvidia’s roadmap seriously, the importance of memory is only going to increase meaning that high-bandwidth memory (HBM) is going to continue to grow like wildfire.
• The result of Samsung messing this up has been poor performance for some considerable time and a share price that fell by over 40%.
• With a discount of this size, the market is assuming that Samsung is not coming back in HBM, but we have seen this sort of thing before.
• When the Note 7 started catching fire in people’s pockets the view was that this would cost Samsung its leadership in smartphones, but the company buckled down, found its backbone and dug itself out of its hole.
• There is every indication that this is exactly what is going to happen here as the problems it has had in HBM3 are fixable and I think that Samsung has the depth of character to do what it needs to get HBM4 right.
• Hence, I don’t think that Samsung is going to get very far with its 12-layer HBM3E but I expect it to qualify with Nvidia for HBM4 that will be going into the Rubin chip that will become available in H2 2026.
• Samsung is currently on 13.2x 2025 and 9.7x 2026 PER and I suspect that the 2026 estimate is too low.
• This sets the scene for a big rally triggered by the news of qualifying with Nvidia for HBM4 which I am hopeful will come this year.
• I have a position in Samsung where I am looking for around $1,600 on the global depositary receipt (GDR) that trades on the London Stock Exchange.

• OpenAI has capitalised on investors’ desperation to invest by raising $6.6bn at a post-money valuation of $157bn and setting very unusual terms that are yet another indication that AI remains in the midst of its 4^th and largest bubble.
• In a short press release, OpenAI announced its fundraising and also went on to give a glimpse of its fundamentals which give more questions than answers.
• Details are as follows:
  • First, free users: of which there are now 250m weekly active users up from 100m weekly active users one year ago.
  • There is some confusion around this number as in January 2023 the company said that it has 100m monthly active users which is a different metric entirely.
  • RFM’s rule of thumb that it has used for years is that 100m users equate roughly to 30m daily active users.
  • Applying this to the January 2023 figure would roughly give around 50m weekly active users as of January 2023.
  • This implies that the user base has roughly doubled every year for the last 2 years.
  • These users are not monetised and generate no revenues and so, in financial terms, they represent -100% gross margin for the company.
  • Second, Value per user: which I struggle to make sense of.
  • At a valuation of $29bn in January 2023, investors were paying $580 per user which rose to $800 per user in early 2024 and now stands at a relatively more reasonable $628.
  • This implies that the present value of future cash flows per user is $628m which is a very tough pill to swallow when most users generate 100% negative gross margin.
  • Second, paid users: According to CNBC (reasonably reliable), there are 11m paid ChatGPT+ users and 1m paying business users.
  • At $20 per month and $30 per month respectively, this gives monthly revenues of $250m ($3bn annualised) which is within spitting distance of the $3.7bn forecast for 2024.
  • However, the expectations are that revenues will grow to $11.6bn next year which will require a subscriber base of 48m more than 4x what it is today.
  • This is supposed to be achieved with no price erosion despite an increasingly bewildering number of competitors and Mark Zuckerberg’s contention that everyone is already “slashing prices” to compete with Llama 3 priced at $0 per month.
• OpenAI expects to lose around $5bn this year (explaining why it needs more money) meaning that total costs are around $8.7bn.
• By my calculations, at $20 per month and 11m paid subscribers, ChatGPT+ should be hugely profitable by itself leading me to wonder where the rest of the money is going.
• This does not seem to matter and demand for OpenAI equity is so strong that it feels that it is in a position to demand that its participants in this round do not invest in its competitors.
• This sounds like an anticompetitive action to me which is the last thing OpenAI needs to do given that regulators are already sniffing around its activities and its business.
• It also appears that OpenAI has been able to backtrack somewhat on its promises to become a proper for-profit company.
• A couple of weeks ago (see here), it appeared that this was a condition of the transaction but this appears to have been downgraded to “we will consider it”.
• The current corporate structure contains conflicts of interest which I think remain large enough to cause the company to implode as it almost did about 1 year ago.
• Last time around the conflicts were fudged rather than fixed and so the risk of an event triggering another existential crisis at the company remains as large today as it was before.
• This kind of conflict has no business being present in a company with a $157bn valuation and the fact that the current investors are willing to simply overlook this issue shows just how desperate they are to invest.
• The one cool head here is Apple which according to reports has declined to invest in OpenAI which is also an indication that Apple Intelligence has no intention of remaining exclusive to ChatGPT.
• I suspect that a large part of the $6.6bn raised this week will find its way into Nvidia’s bank account underlining yet again that it is the only rational way to invest in generative AI directly.
• However, I continue to prefer the adjacencies of inference at the edge (Qualcomm) and nuclear power as their valuations are far more reasonable and both will still perform well even if generative AI fails to live up to expectations.
• As for OpenAI and its competitors, their valuations will only be able to defy gravity for so long and when the correction comes, those that have flown the highest will have the furthest to fall.
• Adding this to the conflicts of interest and governance issues of OpenAI leads me to think that OpenAI will fall the furthest and the hardest when the bubble eventually does burst.

• Meta showcased the results of billions of dollars of investment and at the same time was not shy of taking shots at its competitors as it staked its claim to the next generation of the digital ecosystem.
• Connect is Meta’s showcase for developers where the focus this year was all about AI and The Metaverse (something most people have completely forgotten about).
• In my opinion, ignoring The Metaverse is not a smart move as one day, the technology is going to be ready by which time those who have chucked it on the scrap heap will be too late.
• This is why Apple has the best VR device with the best reality detection system that hardly anyone has purchased because if Meta stumbles upon the formula that popularises The Metaverse, it will be ready to compete.
• There were three main announcements:
  • First, Orion: which is a pair of augmented reality spectacles that are far from ready for the market, but make a number of key advances.
  • Orion has a 70-degree field of view which is pretty good as well as a bright display and all of the compute onboard the device.
  • The demonstration of the device is pretty good but most of the functionality makes use of AI to identify objects and situations in the real world and make suggestions based on that.
  • How much of that was happening on the glasses themselves, remains to be seen.
  • It will also prioritise voice, but gestures and neural impulses are also being worked on.
  • The demonstrations were slightly glitchy but by and large, Orion is a good step forward but remains far from anything that a consumer is going to use on a daily basis.
  • Meta has shown its progress, and the results show that if The Metaverse takes off, it will be in pole position to be a major player and not have to remain dependent on Apple or Google.
  • This has been one of Mr Zuckerberg’s gripes for many years and ever since Apple messed around with his business model, he has been adamant that this will not be repeated.
  • Orion is a step along that journey but the road yet to be travelled remains long.
  • Second, AI: where Llama 3.2 was released in its 11bn and 90bn parameter sizes but not the really big 405bn variant.
  • Llama 3.2 is multimodal for the first time and catches up with the latest versions of Gemini and OpenAI in terms of capabilities.
  • Here, Mr Zuckerberg could not resist taking a swipe at his competitors stating that Open Source is the way to go and that his competitors are already slashing prices to stay competitive with Llama.
  • I have seen no hard evidence of this but if it is true, it is a sign that the bubble is beginning to burst.
  • Llama 3.2 11bn and 90bn will not be available in EU and Mr Zuckerberg once again could not resist taking a shot at the EU for its excessive regulation.
  • I continue to think that with its flagship 405bn Llama 3.1 model available for free to anyone that wants it, Meta is rapidly commoditising generative AI and its foundation models meaning that differentiation will increasingly be in the services that are built on top (see here).
  • Third, VR: which is where the volume of The Metaverse exists today.
  • A new, cheaper device Quest 3S was launched which starts at $299 which is clearly a subsidised price.
  • This device is all about a land grab and ensuring that all early moves in The Metaverse are made using Meta’s platform and with its independent competitors having largely given up, the road ahead is clear.
  • This will light no fires under Metaverse adoption but crucially it will ensure that Meta dominates the fledgling market positioning it well for a take-off if it comes.
• These launches are all about positioning Meta well in the growing market for AI services and The Metaverse when / if it takes off.
• Meta was not shy in calling out the shortcomings of its competitors and regulators and, to be fair to Mr Zuckerberg, he is putting his money where his mouth is and delivering progress.
• Llama is already a thorn in the side of Google, OpenAI, Microsoft and Amazon and is building a barrier to entry for Apple in both AI and The Metaverse.
• 5 years ago, Meta Platforms was a laggard in AI and while it has spent a fortune, it has gone up significantly in my rankings and is now challenging the leaders.
• However, it may also be the cause of the bubble bursting and so while there is AI upside priced into Meta’s shares, I would be wary of investing.
• The cheapest AI investment available remains Baidu which along with its peers is having a bounce as China has finally decided that it needs to do something about its moribund economy.
• It is still risky, but this could be the beginning of the long-awaited rerating of Chinese technology.

• It is said that a company is more than just one man but when almost all of the brains of the company are no longer there, one has to wonder what this means for product launches, revenues and of course, value.
• In a tweet clearly drafted by a lawyer, Mira Murati, CTO of OpenAI has announced that she is leaving the company which in my opinion is a heavy blow to OpenAI.
• This is because she was the brains of the operation and the additional departure of Chief Research Officer, Bob McGrew and Barret Zoph, VP of research compounds my concerns even further.
• Furthermore, despite plenty of hyperbolic commentary, GPT-5 is nowhere to be seen, and Sora (video generation) still has not shipped despite being demonstrated in February of this year.
• There are two possibilities for these departures.
  • First, Start-up where Mira and her colleagues found their own company and create a competing product.
  • This is happening with increasing frequency and there are now many start-ups headlining with alumni from OpenAI and Deep Mind.
  • Given how important Mira was to OpenAI and how frothy the market is, I suspect that she will have no difficulty raising money for her new venture if that is the path she has chosen to take.
  • Second, disagreement: which has also been fairly common where founders reach an impasse on the direction that the company should take and part ways.
  • Ilya Sutskever, one of the other main founders of OpenAI also recently left for this exact reason and has started his own company focusing on AI safety as this is where his opinions and those of Sam Altman diverged.
• On top of the fundamental implications of Murati’s departure, the timing could not be worse.
• OpenAI is currently raising money at a pre-money valuation of $150bn putting the company on 30x 2025 revenues in an optimistic scenario.
• This assumes that the company will become a proper for-profit (highly likely in my view see here) and that revenues grow very quickly and that fat profits will follow.
• With the brains of the operation gone and possibly gearing up to compete with OpenAI directly and rapid growth in competing offerings, this proposition is becoming more fanciful with every passing event.
• OpenAI is no longer miles ahead of the competition and with Meta’s party-blowing release of its flagship model and its weights into the open-source community, the stage is set for rapid price erosion.
• There are already signs of this as more features are making their way into free generative AI services, but it is not until prices begin to fall that anyone will notice.
• This is the pin that bursts the bubble in my opinion and given how much it has been inflated, there will be a significant correction to reality.
• This is why I remain very nervous about the valuations that are being paid for these companies and am staying well away from them.
• If I were forced to invest in this area, it would be Nvidia which actually has revenues and profits now or Qualcomm which is in a very good position to benefit as generative AI starts to be implemented at the edge.
• I already own Qualcomm which remains reasonably valued and has further upside from its position in automotive as well as the potential to take share from Intel and AMD in laptop processors.

• OpenAI’s corporate structure has been unfit for purpose for over a year now but finally, it looks as if investors paying a pre-money valuation of $150bn are forcing OpenAI to become a proper company rather than a science project with a commercial attachment.
• OpenAI as it exists today is a powder keg that has almost killed the company once and because the structure was fudged rather than fixed, it could blow up again at any moment.
• OpenAI was set up as a non-profit to develop artificial general intelligence (AGI) which is the point at which machines become as intelligent as humans or more so.
• As a non-profit, OpenAI would ensure that this critical technology would not be monopolised by a single company but would instead benefit all of humanity.
• While the company was a bunch of scientists tinkering with a problem that has eluded humanity for years and making little progress, this was not a problem.
• However, along the way on this journey, the company created something that does not solve the AGI problem but did cost a fortune to create and has very great commercial potential.
• When OpenAI took $1bn from Microsoft in 2020 which it needed to pay for the immense amount of compute that was required to develop its large language models (LLM), the seed was sown for potentially fatal internal conflict.
• The problem was (and remains today) that the company is governed as a non-profit but with more and more of the company being owned by commercial investors needing to see returns the pressure to make money continues to increase,
• This was compounded in 2023 when OpenAI took another $10bn from Microsoft and with the current round at $150bn pre-money, the pressure looks like it is becoming unbearable.
• Microsoft is a commercial enterprise and does not invest $11bn with no hope of earning a return as it has a fiduciary duty to its shareholders to make them money.
• This is why I suspect that Sam Altman discussed this at the all-hands meeting held at OpenAI last week as this condition of the round was almost certainly going to leak.
• While he would not be drawn on specifics, it is clear that the company will move away from its non-profit roots and hopefully become a proper for-profit company.
• This would be very good news for everyone concerned as Microsoft is embedding OpenAI technology in all of its revenue-generating products and a further implosion could cause Microsoft real problems.
• While this will greatly reduce the risk attached to OpenAI, it does not suddenly mean that $150bn pre-money represents good value for investors.
• For example, even if OpenAI generates $5bn in revenues next year, it is unlikely to make any money which is a real problem for a company at this scale.
• It raises questions about the financial returns that can be made from generative AI as the costs seem to be going up as fast, if not faster than revenues.
• Furthermore, the idea of paying 30x revenues for a company that doesn’t make money and is not a tiny start-up just beginning to make revenue makes no sense to me.
• However, the money is almost certain to generate more demand for Nvidia which remains the safest way to play the generative AI craze directly.
• However, I continue to prefer the adjacencies of inference at the edge (Qualcomm) and nuclear power as their valuations are far more reasonable and both will still perform well even if generative AI fails to live up to expectations.

• Inviting Apple as an observer onto its board is a pitch to Apple to more deeply integrate OpenAI into its ecosystem but given how unstable OpenAI is, I expect that Apple will continue to decline its advances.
• In the run-up to WWDC, there was a lot of chatter about how Apple was going to integrate OpenAI into its ecosystem such that when the reality was revealed there was a lot of misunderstanding.
• Apple Intelligence (see here) is a completely self-contained system which as far as I can see, includes no OpenAI technology at all.
• Instead, Apple offers the option to send the inquiry to one of the large language models (LLMs) available on the web of which OpenAI’s ChatGPT is the first to be enabled.
• Given the hype prior to WWDC, this has been misinterpreted as deep integration into Apple Intelligence which is clearly not the case.
• This is evidenced by the fact that Apple is going to enable the other LLM providers in time and so users will be able to choose which one they want to use very much like search.
• However, this move by OpenAI to invite Apple onto its board leads me to think that OpenAI was pitching Apple for deeper integration into Apple Intelligence which Apple declined.
• I think that Apple declined OpenAI’s advances for three reasons.
  • First, Not made here: Apple likes to develop its own technology where it can and not create the kind of dependencies that others have on Apple.
  • Hence, allowing 3^rd party AI into Apple Intelligence has the potential to create a dependency that Apple thinks that it does not need.
  • Second, Small Language Models: What little we know about Apple Intelligence is that the on-device system is a lot like CoPilot+ which is a runtime full of small models that can each perform a generative AI task.
  • OpenAI does not address this requirement as it firmly believes in ever-larger models consuming more data and more computing power.
  • Hence, I don’t think OpenAI had anything to offer Apple for the Apple Intelligence piece that it would want.
  • Third, Basket case: OpenAI has demonstrated that it is a powder keg waiting to explode.
  • Despite almost imploding in November of last year (see here), it has not addressed the fundamental issues and simply papered over the cracks meaning that it could just be a matter of time before it goes off again.
  • Hence, the last thing any prudent user of OpenAI technology should do is to create a dependency on the company as a source of generative AI.
  • This is where I think Satya Nadella has made pretty much the only mistake of his tenure at Microsoft although he does have the option to acquire the company should it go bang again.
• Hence, I am not surprised at all that Apple declined to integrate OpenAI more deeply into its digital ecosystem and I can completely understand why it will continue to try and convince Apple.
• Google makes a substantial portion of its revenue from the Apple ecosystem and even though it pays many billions of dollars for the being on the ecosystem, it remains extremely profitable.
• Hence, it makes complete sense from a commercial perspective that OpenAI would seek to be integrated into the Apple ecosystem.
• However, I suspect that once Apple gets a look under the hood by sitting on the board meetings, it will decide that OpenAI needs to be kept at arms-length given the corporate governance issues and the risk they represent to anyone who has a dependency on OpenAI.
• I also suspect it will also conclude that the capability of its models is broadly in line with everyone else’s and that there is no advantage in having an exclusive relationship.
• The net result is that I don’t think that this signals a deepening of the relationship between Apple and OpenAI and in fact, may reinforce Apple’s decision to remain at arm’s length.
• The inevitable implosion of OpenAI is why I think that Microsoft has no choice but to buy it at some point.
• This point is likely to be either the next time OpenAI sets itself on fire or when the AI bubble pops and everyone loses interest and cuts spending.
• This is why when I think about the competition in AI, I think of Microsoft and OpenAI as one company.
• This acquisition looks inevitable to me.

• 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.

• Google continues to struggle with its messaging as OpenAI was able to deliver more in 30 minutes than Google did in a long, rambling and discombobulated 2 hours.
• If Google intended to amaze and astound then it failed badly, but it did manage to deliver a solid set of updates that will ensure that its ecosystem services remain the best available which is crucial as these are the engine of its revenue generation.
  • First, numerical updates: Here, Google stated that it has 1m developers using Gemini in some way or another which is great but then went on to say that only 1m users had signed up to use Gemini which is pretty poor.
  • However, this excludes the billions of users who have Android phones where Google has an opportunity to reach them directly by updating the Android OS they are using.
  • Google did not announce a huge new model but did expand its lead in context in terms of increasing the context limit to 2m tokens from the 1m where it is today.
  • This is important for the heavy enterprise use cases but for the average digital ecosystem user, this will make no difference whatsoever.
  • Second user experience: which OpenAI put front and centre, but Google was much less clear on.
  • Here, Google is introducing a new model called Gemini 1.5 Flash which still runs in the cloud but has been optimised for speed as Google has recognised that responsiveness is going to be key going forward (see below: voice).
  • Google will also update its on-device model, Gemini Nano, so that it is properly multimodal in terms of input capability as well as more responsive thanks to being on device.
  • Third, Voice: which was an important part of all of the demos but nowhere did Google really push voice as being an important aspect of the user experience as OpenAI did.
  • Google introduced the ability to interrupt the model which adds to the natural experience, but because OpenAI introduced it first, it will get credit for it even if it borrowed the idea from Google in the first place.
  • The main takeaway here is that two of the big AI providers think that voice is going to be important which is something with which I completely agree.
  • The best use case for a system that is good at natural language is to use it as a man-machine interface and RFM Research has long identified this as one of the best use cases for LLMs.
  • Fourth, Open Source: Driven by Meta, Google is being forced into the open source where it has a 2bn and 7bn variant of Gemini available which it calls Gemma.
  • Google is updating its offering here and will also add a 27bn variant in response to developer requests, but I have doubts about whether it will displace Llama which is already the standard for open-source generative AI development.
  • Fifth, Android: which for the first time ever pushed the actual updates to the OS into day 2 and instead focused on Gemini’s integration into the OS.
  • There are some interesting advantages here as Gemini can now be contextually aware making using it in the context of what the user is doing easier and more natural.
  • It is also a point of differentiation over iOS where Apple still has nothing outside of the execrable Siri but is expected to announce something with OpenAI at WWDC in June.
  • Google is expecting to have Gemini embedded into 200m devices by the end of 2024 in a sign of just how powerful controlling the platform can be.
  • This is Google’s best shot at dominating the AI ecosystem because if users start to use Gemini and like it, then developers will be more inclined to develop to Gemini as opposed to GPT or any of the other foundation models.
• When one looks under the surface of this disjointed performance, it is clear that Google is one of the leaders in AI and given the ecosystem that it already has, it has the best shot at dominating this space.
• However, OpenAI and Microsoft are now really good at getting under Google’s skin and making the most of the fact that Google is an engineering-led organisation which is not very good at marketing its own prowess.
• This alliance is clear and explains why OpenAI only launched ChatGPT for the Mac leaving the space open for Microsoft’s Copilot which is based on the same technology.
• Despite another clumsy performance, Google has a strong position in the AI ecosystem claiming that 1m developers are already using Gemini.
• Google did not clarify how many of these are actually building generative AI on Gemini as opposed to using it to write regular apps but either way, it’s a good number.
• This puts Google ahead in terms of the race for the AI ecosystem, but OpenAI is punching well above its weight and has global user recognition of the ChatGPT brand also putting it in a strong position.
• OpenAI and Microsoft have also made Google look like a laggard on numerous occasions and have done so yet again this week, but this is just the first battle in a war that will last for several years.,
• Microsoft and Apple have yet to lay out their wares for generative AI but I/O evidenced good and steady progress at Google even if it remains very poor at getting the message across.