Robert Paul Leitao on The AI burn rate is making Washington nervous - 'I advocate for learning and informed discourse about most issues of consequence. On this issue, I am a big proponent of AI and its positive impact on our economy and on our society. Already I see the positive impact is it having on Apple’s prospects for continued success despite the near-term challenges of component supply constraints due to AI infrastructure demand. There will be both winners and losers in the AI race and I do a lot of research to determine in which AI related enterprises I will make investments. I have several AI-related holdings at this time. Among the various professional positions I have held over my career several have been related to finance and accounting including audit preparation and a general understanding of accounting rules – both FASB and GAAP, and rules compliance. I’m impressed with the results both Microsoft and Amazon just delivered and these large enterprises are certainly following established accounting rules and public audit procedures. I’m also interested in what Alphabet is bringing to market and the resources from the company Apple has chosen to deploy. I’m fascinated and excited about what’s occurring in the marketplace today. Among the reasons I bring this up is because Apple is also making some big commitments with suppliers. Recently the company signed a $30 billion+ agreement with Broadcom and Apple has committed to invest $600 billion in the US economy over the next four years. Neither of these commitments are reflected directly on the company’s balance sheet. Apple is hiding nothing. After reading about the new $30 billion+ agreement between Apple and Broadcom I took a look at Broadcom from an investor’s perspective. The company’s shares closed on Friday at $389.28. In a note dated July 8th Morningstar has a $650 Fair Value Estimate on the stock and in the note the deal with Apple is highlighted under the title, “Broadcom: One More Bite at the Apple”. The agreement with Broadcom is intended to boost American chip production and is part and parcel of Apple’s $600 billion investment in the US over four years. I know there are a lot unknowns about AI implementation and inculcation into our society. For every concern there might also be an opportunity. I’m not making a stock recommendation or urging or advising anyone to purchase Broadcom shares. Please perform your own due diligence and act with knowledge, wisdom, insight and discretion. I’m just making a point that I also see opportunities while we all consider the uncertainties and perceived risks. I remain bullish on Apple and I’m looking forward to the opportunities being created in this AI era. Go Apple! Go Apple 3.0! May we all live long and prosper!'
on The AI burn rate is making Washington nervous - 'Last year everyone was hung-ho on AI. Now it has flipped. Even Jim Cramer is worried about CAP-EX spending. The big problem is that there doesn’t seem to be a way for the big spenders to make a profit. This can’t end well.'
on Apple's mixed Q3 2026: What the talking heads are saying - 'Kramer from Arete makes some very good points: 1. Apple’s margins held up reasonably well up 60 basis points YoY when tariffs were stripped out, while Samsung’s margins, arguably the best in Android, fell by 12%, percent, not basis points. 2. Apple is providing very conservative, almost pessimistic guidance for Q4 and some color for Q1. Kramer feels Apple may be “de-risking” the quarter and guide analysts to a beatable number instead of a rosy picture. I’m not sure if Apple could be accused of sandbagging, because the memory cost issues and Fx headwinds remain quite real, but it’s an interesting thought that Apple wants to set the stage for a good Q4 earnings report following the iPhone 18 Pro and Ultra introductions, among other products. 3. $50 billion worth of Apple Silicon AI capable chips is already out in existing devices when Siri AI and Apple Intelligence comes online with iOS 27. And all newest iPhones will also be capable, expanding that population consistently. 4. All of that Edge AI capability was completely funded by users, Apple’s costs were the R&D and production costs only, already covered by iPhone sales. 5. Kramer claims Apple iPhones rule the refurbished smartphone market at 90%, probably pointing to the US market. Checking with Gemini reveals a. Apple has a 58% share of global refurbished sales. b. In the US, Apple has ~40-45% of third party refurbished sales. • Separately, carrier based trade-in and end of contract supply and refurbished models show iPhones command 65-70% of Carrier refurbished sales. c. For India, rapid turnover and high demand have given Apple a 64% share of the refinished market. • for India and likely repeated in other markets: —> Driven by high inflation and recent price hikes in the new sub-$200 budget tier, Indian buyers are actively shifting away from cheap new phones toward higher-tier refurbished devices. The Shift to Premium & Pro: • For years, basic base models like the refurbished iPhone 11 and iPhone 12 dominated Indian volume. However, Counterpoint Research highlights a massive structural pivot toward Pro and Pro Max variants (such as the iPhone 13 Pro and iPhone 14 Pro), where consumers feel they get the most significant “status symbol” value for their money. • Android Flagships: In the Android space, the most popular refurbished choices are Samsung Galaxy S21 and S22 series devices, alongside premium Google Pixel alternatives.“ d. Back to the US 1. The Value vs. Volume Split • By Unit Volume (~40% – 45%): In terms of the actual physical number of phones shipped through US third-party resellers, iPhones make up roughly 40% to 45% of total units. Android devices collectively capture a higher unit volume (around 50%) because the market is flooded with massive numbers of cheap, lower-end, pre-owned Android models. • By Total Revenue (>60% – 70%): Because iPhones cost significantly more on the secondary market than the average Android phone, they capture the lion’s share of the money flowing through the industry. A reseller might move three low-end Android phones for every one iPhone, but that single iPhone can generate more revenue than all three Android devices combined. 2. The Apple Depreciation Advantage The core economic driver behind the 60%-70% revenue dominance is value retention. According to resale market data from platforms like SellCell, iPhones depreciate significantly slower than their competitors: • iPhones retain roughly 60% to 65% of their original retail value after two full years of use. • Android flagships typically lose value much faster, dropping to roughly 30% of their original value over that same two-year window.This allows third-party resellers to maintain a high average selling price (ASP) on Apple inventory. While the average used Android phone often sells for well under $200, the average selling price of a refurbished iPhone in the US consistently sits much higher—hovering around $380.'
on The AI burn rate is making Washington nervous - 'Per my stepson, who works in AI and computer security, regarding the idea that pouring massive amounts of cash into data centers will eventually produce a high ROI: “It presumes that the largest of models intelligence will be so great that most applications will require it – and those models will require data center capacity. However, even now we can run a 1 trillion parameter model on 4 GB of RAM. It’s slower, but fast enough for local automation workflows. And it’s getting to the point that more parameters are not necessarily giving significantly more performance. Not only that, but plenty of use cases will only require 70 billion parameter model intelligence. I think local inference will be the future. Data center compute is still needed to TRAIN and create new models, but inference for those models can be local. There is a method called distillation where you can take a 1 trillion parameter model and distill it down to 100 billion model, with less than 2% degradation of performance. So, plenty of reasons local inference will be cheaper and better. Especially for 80-90% of the applications. We’re at the point where quality data, architecture, and techniques are the limit of how to improve AI – not size. There is a diminishing rate of return with just going bigger, in terms of AI performance. Like think in terms of CPU: it’s not just number of transistors. It’s also clock speed, bus speed, RAM speed [unified memory, such as is found on Apple products, being quicker], and other aspect that make the application perform faster.”'
on The AI burn rate is making Washington nervous - 'All that Siri came up with for your prompt is: I can’t create an image of that. Try describing something different. While better, it hasn’t been very good for my usages.'
on The AI burn rate is making Washington nervous - 'Ovide writes: “But questions about that AI vision are now growing more urgent: When, if ever, will this payoff arrive?” The answer arrived quarters ago and its confirmation is available in new earnings reports. The accelerated computing data center business model is sound and the ROIC from its infrastructure build is impressive and already clearly underway. There will be failures because always there are failures along the change pathway. But for anyone who wants to evaluate objectively, there is ample evidence this shift in computing platform rapidly is demonstrating its economic viability. The quotes are Andrew Jassy’s. “Let me talk for a second about how we see this investment playing out. Earlier this year, we said we plan to invest approximately $200 billion in cash CapEx in 2026, the majority of which to support AI and AWS. At this level of spend and higher, we have clear line of sight to strong financial returns. I’ll explain why. There are 2 major parts of the investment, the data centers and the servers and networking equipment that go into them. These have different capital cycles. Data center capital is spent starting 2 years before we can put servers into them to start monetizing. Once a data center opens with servers plugged in, we start generating significant revenue right away and then get to monetize these data centers for 30-plus years without having to spend that start-up capital again. Servers and networking equipment operate on a shorter cycle. We typically purchase these a few months before putting them into service, so we have strong visibility into customer demand before we trigger the spend. If the demand isn’t there, we won’t spend the capital. For servers and networking equipment, on average, it takes a little less than 3 years to break even on that investment. The servers currently have a useful life of at least 5 to 6 years, and most of our AI capacity these days is being contracted for at least 5-year terms. That means that we’re driving significant free cash flow on the servers and networking equipment in the 2 to 3 years after we break even. It’s also worth noting that AWS has a strong track record of pulling forward break evens on server equipment where we’ve already made meaningful progress and finding ways to extend the useful life of this equipment without sacrificing customer experience. So for our data centers, which have 30-plus-year useful lives, we should get at least 5 to 6 generations of server economics, like I explained earlier, with subsequent generations after the first having even better overall economics because we don’t have to repeat that upfront data center investment I mentioned earlier. This means in the short term, when demand is necessitating so many data centers being built simultaneously in advance of when we can start monetizing them, we’ll spend a lot of CapEx and encounter free cash flow headwinds until these data centers come online, can be monetized and we get a few years into these servers being utilized. But as we get a few years out and the revenue growth outpaces the incremental CapEx growth, which will happen at some point, the resulting revenue, free cash flow and return on invested capital is very compelling. We’ve done this before in the first era of cloud computing, just over a longer time horizon, where demand built more gradually than it has in AI. But we see the margins and returns in AI tracking what we saw with core at the same point of evolution, actually, a little ahead.”'
on The AI burn rate is making Washington nervous - 'Am I right in believing that SOMEONE is providing the money for all that debt? Or is this a case of paper debt and paper profits?'
on NYT: The day Trump put Larry Ellison in charge of AI - 'Ellison should spend his time racing his yacht. Or watching it race.'
on The AI burn rate is making Washington nervous - 'Hyperscalers are also hiding an additional $1.x trillion of debt from their balance sheets. lycoristechnologies.com/blog/hyperscaler-off-balance-sheet-ai-debt/'
on The AI burn rate is making Washington nervous - 'This is tangentially related to AAPL and stock price outlook. I saw that TD Cowen today issued a buy rating and a $400 price target on AAPL. That’s optimistic I think. I guess we wait to see what Dan Ives has to say.'
on WSJ: Apple gave $25 million for Trump's ballroom - 'It’s impossible to reconcile Cook’s history with his willingness to play along with Trump. Just a few examples: In 2014, Cook told shareholders who didn’t agree with Apple’s position on climate change to “get out of the stock.” In 2017, Cook said Trump’s DACA stance was un-American. “This is unacceptable.This is not who we are as a country. I am personally shocked that there is even a discussion of this.” The Robert & Ethel Kennedy Human Rights Center may want its “Ripple of Hope” award back. The award “recognizes leaders who demonstrate a dedication to improving the world, reflecting Robert Kennedy’s values.” https://kennedyhumanrights.org/press/john-lewis-tim-cook-roger-altman-marianna-vardinoyannis-win-2015-ripple-of-hope-award/'
on The AI burn rate is making Washington nervous - 'Here is a nice breakdown discussing how debt impacts much of the AI economy — the recent burst of the Situational Awareness AI fund, the circular funding, the impact of AI on our GDP and even the recent Fed decision — possibly leading to a market meltdown similar to 2008. For many of the same reasons most bubbles burst — margin calls and loan defaults. I may not always agree with Max, but I often do learn something new. He usually has some facts and graphs on topics I am not aware of. Situational Awareness: How a 25-Year-Old’s Hedge Fund Exposed the Entire AI Bubble'
on NYT: The day Trump put Larry Ellison in charge of AI - 'His immediate willingness to throw Elon Musk a billion or more $$$ to buy Twitter was part of all of the influence game required to do what he has been in AI in TrumpLand.'
on WSJ: Apple gave $25 million for Trump's ballroom - 'All they need is a TV with Fox News 24/7, a phone to post unhinged rants and call people to yell about fraud, Dems how he is the greatest _____ and has been wronged.'
