DoorDash just told you what it thinks Uber is worth. It's buying Instacart for $65 a share, about $17B and at least 75% in cash. That works out to about 13x Instacart's free cash flow. Uber trades at roughly the same, about 13x. DoorDash itself trades at over 30x. So the market prices Uber like a grocery app that just got bought out, while the buyer sits at more than double that. Uber was in talks for Instacart and walked away. My guess is that the price was the main issue. Why should Uber pay a premium for a business growing slower than its own grocery arm. Uber's grocery and retail business is growing over 40%. Instacart's order value grew 15% last quarter. CFO said that each transaction is compared to their own buybacks among other things and so I think price could really be an issue in this case for Uber. I also said the real risk was DoorDash. Now it's here. Put DoorDash and Instacart together and they own the large majority of US grocery delivery, with Uber a distant No. 2. DoorDash already leads in restaurants. And Instacart's restaurant tab runs on Uber Eats today. After a deal, some volumes will disappear. But it won't happen fast. The deal isn't expected to close until the end of 2027 or the first half of 2028, and antitrust is the big question. The last time these two talked, in 2021, antitrust fears were part of why it died. That's 14 to 20 months where grocers and advertisers don't know how this ends, and some will want a second option. Uber is the obvious one. Uber reports Wednesday, Nov 4, before the open. I'm sure someone will ask about Uber walking away, whether grocery is still growing over 40%, and ultimately whether buybacks are picking back up. At about 13x free cash flow, Uber is priced like the company that just got bought. To me, it’s the stronger business: global reach, a broader revenue mix, and faster growth. If you want more updates like this, follow me @m0xt_ for more.

ServiceNow is down about 23% after a quarter that was fine. The likely reason is not the quarter, but the guidance. Subscription revenue grew 21%. Contracted revenue for the next 12 months, the north star, beat its own guidance. But next quarter's guidance for that same number is 18.5%, and the Street wanted closer to 20%. It gets worse once you strip out Armis, the security company ServiceNow just bought. Without it, growth is closer to 17%. Look at the chart. That number peaked at 23.5% two years ago, then sat around 21% for two years. Now the guide breaks below that range for the first time. And the beat wasn't as clean as it looked. Management said about half of it came from customers renewing early, so that's revenue pulled forward from next quarter. Then there's AI. The AI business keeps growing, and management raised this year's target again to $1.7B. But total growth is slowing anyway. This is the part I keep coming back to. ServiceNow charges per seat, meaning per human using the software. If AI agents do more of that work, there are fewer seats to sell. I think the whole business has to go through a full transition. Away from seats, toward pricing based on tokens/queries or outcomes. Every time someone asks a question, ServiceNow's own model answers it, and it gets paid for that answer, more like tokens than licenses. Management says customers aren't asking for outcomes-based pricing yet. I don't think seat pricing lasts much longer anyway. And a transition like that is rarely smooth for revenue and margins while it happens. The market is already pricing a lot of that in. Free cash flow per share went from $1.77 in 2021 to about $4.55 today, while the multiple fell from about 71x to about 26x. On the next 12 months of earnings, it trades at roughly 25 to 27x, roughly where Microsoft trades today. Banks are cutting targets but mostly staying bullish. Morgan Stanley went to $160 and Truist to $150. But the risks are becoming clearer, too. Software is getting derated across the board. And if ServiceNow turns into a 15% grower, the cheaper software names show where the multiple can go. I exited my entire position around $135 since the coming quarters look challenging for ServiceNow as they adjust to the new reality. They’re still well set up to stay important, but I wouldn’t be surprised to see the valuation multiple get compressed. Follow me @m0xt_ for more market updates.

Morgan Stanley thinks US data centers will be about 43 GW short of power through 2028. And that's after every fix it could find. Look at the chart. US data centers need about 116 GW from 2026 to 2028. Take out what's already being built and what the grid can deliver, and you're still 70 GW short. Then Morgan Stanley subtracts nuclear, new gas turbines and engines, and fuel cells. Still 43 GW short. Even if 5 GW of that demand moves to data centers in Asia, the gap is 38 GW. That's a third of everything needed. In September, the same estimate was about 33 GW. Part of the jump is method, not new orders: the bank now counts more of each server rack beyond the GPUs, plus more old non-AI demand. But the gap has stayed somewhere between the low 30s and the low 40s all year, even as the bank keeps adding fixes. Grid queues are full of speculative requests. One Ohio utility's queue fell from 30 GW to under 6 GW once it made applicants commit. Chips also keep getting more efficient. Morgan Stanley itself expects tokens per watt to rise about 6x from 2025 to 2028. And a shortfall doesn't mean the demand disappears. Some of it just slips into 2029 or moves abroad. Still, power remains the scarce asset, and whoever already has it switched on sets the price. The industry refers to these firms as Powered Shell Providers (PSPs), and I dig into one of them here: youtube.com/watch?v=EcJ-6OKN… PSPs are my second biggest allocation on Milk Road. Go and check it out: link.milkroad.com/Martin

SK Hynix could buy back about 10% of its shares in 2027, and about 15% in 2028. You really think a $250B buyback won’t move the stock meaningfully? That’s your window, and it won’t stay open indefinitely. The market still prices SK Hynix like a commodity chip maker. On Oct 28, SK hynix said it will return about 100T won to shareholders for 2026. At least 30T in dividends and at least 60T in buybacks, and the shares bought back for shareholders get canceled. That 60T is about 5.5% of the company at today's price. The bigger number comes after. The policy is to return over 50% of cumulative free cash flow from 2025 to 2027. Dividends get paid first, and the rest goes to buybacks. So I ran it on 9 broker reports published the day after results. Leaving out the highest and lowest, their 2027 free cash flow estimates run from 242T to 357T won. Take half, keep paying the regular 10,000 won a quarter, and the buyback pool lands at 93T to 151T won. At today's price, that's 8.6% to 13.9% of the company in one year, with the middle estimate at 10.8%. That middle estimate is about 118T won, roughly $88 billion of buybacks in a single year. If the policy gets extended into 2028 at the same 50%, it's 146T to 182T, or 13.4% to 16.7%. The middle there is about 164T won, roughly $122 billion. Brokers are saying it outright. SK Securities pencils in 158T won of shareholder returns in 2027 and 205T in 2028. Nomura says buybacks through 2028 could retire about a third of the shares. The most bearish broker gets you just 5%. But the middle of the pack is a memory maker, buying back about a tenth of itself in a year. The stock sits about 50% below its 52-week high, after an 83T won operating profit quarter at an 80% margin. Memory is the most profitable layer of AI right now, and it's priced as if its pricing power will disappear tomorrow. This matters even more for SK Square. SK Square owns about 20% of SK hynix, and that stake alone is worth about 1.79x SK Square's entire market cap. Every canceled share makes that stake bigger without SK Square spending a won. The 40T won program from August is done, and those shares get canceled on Nov 5. That alone lifts SK Square from 20.00% to about 20.67%. This stock is engineered to climb, and I’m glad to be along for the ride. Memory takes up the biggest slice of my portfolio. Check out the rest here: link.milkroad.com/Martin

Galaxy's portfolio book is your best shot at crypto adoption, and I don't think anyone will beat them in the long run. Many people, me included, are increasingly bullish on institutional adoption. But that's not an easy bet because buying coins doesn't really get you there. So here's what I keep asking myself: Is $GLXY a better bet on crypto winning than picking crypto yourself? I tend to think so, because it combines the best of it all. For years, the crypto portfolio has done the heavy lifting in their top-line performance, and it will likely keep that crown until Helios Phase II goes live in Q2 next year. Either way, what matters is that their crypto exposure becomes more compelling quarter after quarter, and I expect their investment lineup will be tough to top. Here is the case: Galaxy trades, lends, builds, manages money, stakes, advises, and invests. Each piece sees something different. The trading desk works with 1,763 counterparties, so it knows what institutions want and what they're hedging. The lending book averaged $1.8B last quarter, up 27%, so it knows who wants leverage and against what. Asset management and staking manage $5.9B in funds and $4.6B in staked crypto, so they know what funds and ETF issuers want to own. But Galaxy keeps building on top of that too. SWEEP. A tokenized Treasury cash fund with State Street, built on Galaxy's tech. Stablecoin holders park cash, earn yield and can pull it out 24/7. It's close to $300M already, and Galaxy's own trading and lending desks take it as collateral. GOFR. One onchain lending rate across Aave, Morpho, Spark, Kamino and others, published daily as a public reference point. Institutions borrow from Galaxy instead of going into DeFi themselves, and Galaxy puts up to $100M of its own money in first loss. It's a big part of why the loan book grew. The yield side. An onchain yield fund with Sharplink with $125M committed, Curator vaults sold to 2,400+ institutional clients through Fireblocks, and Premium Yield, an 8% note for accredited investors in the GalaxyOne app, paid from the lending business. See the pattern? Every product feeds another business. The cash fund becomes collateral for the desks. The lending rate grows the loan book. The loan book pays the retail yield. And this was my favorite part of Tuesday's call: Galaxy has become a core builder too. Some big institutions want to own the core pieces themselves, like custody. So Galaxy helps them build it and gets paid upfront. Then it sells the same platform as a recurring service to everyone else, down to small and mid-sized firms. BNY is the clearest example. Galaxy is its custody client, staking vendor and design partner all at once. Sharplink got a fund set up just for it. ETF providers moved over $400M of new stake to Galaxy in Q3. And Novogratz said Galaxy is in talks with a few other large players. The real prize is what comes after the first deal. Once Galaxy is the partner, the trading, the lending and the next product come with it. A competitor that only does custody or only does trading can't copy that by hiring one more desk. The bull case for their crypto biz is quite obvious and hard to argue against. A reasonable counterpoint is that many of these offerings are still tiny, or have not even revealed their scale. That is painful for short-term traders, but largely irrelevant for long-term holders. In Q3, the crypto segment still posted an adjusted EBITDA loss of $9M. Strip out a one-time accounting adjustment, though, and the crypto segment would already be in the black. The engine is assembled. For now, it’s only beginning to show up in the numbers. As Novogratz put it: slowly but surely. Then it will explode. This post isn't about the crypto business; it's about whether you think this company, which sees every piece in that market, will not deliver outsized returns. I think they will. All three of Galaxy’s cylinders are on the verge of exploding within the next 12 to 24 months. That's why GLXY is my biggest position: link.milkroad.com/Martin

Microsoft guided to over $50B in capex last quarter. It spent $45.2B. Is the AI boom finally slowing? No, it is not. The money was ready. The capacity it rented externally arrived late. CFO Amy Hood: "Capital expenditures were $45.2 billion as some third-party capacity deliveries we expected in Q1 were delayed." The Street was at about $50.7B. Hood also trimmed 2026 capex from about $175B to about $170B and pinned the whole cut on those same delays. Same call: "Customer demand continues to exceed available supply." Azure still grew 48%. Then Google reported the same evening, saying the same thing. CFO Anat Ashkenazi: "Given the supply-constrained environment, we started utilizing third-party compute capacity at the end of Q3 and will ramp usage in Q4." This is a company that spent $51.4B on capex in one quarter, kept its $195-205B plan for 2026, and says 2027 will "increase significantly". It still has to rent capacity outside its own. Google Cloud boss Thomas Kurian: "we could obviously serve a lot more demand with capacity." So two of the biggest AI builders are saying the same thing. They can't pour concrete fast enough, and the gap goes to whoever already has racked and powered GPUs. That's the neoclouds. They're the pressure valve. Now look at premarket. CoreWeave and Nebius are down about 6%. The market is still punishing suppliers after the world’s biggest companies signaled they will rely more on neoclouds than expected. That shift gives neoclouds room to charge steep premiums, cash that could help them navigate a tighter credit market. These two earnings calls were constructive for the broader AI trade, particularly for neoclouds. Neoclouds offer the capacity and product that hyperscalers cannot supply in sufficient volume today and may not be able to supply anytime soon.

Every time a neocloud's debt wobbles, Galaxy gets sold like it borrowed the money itself. I talked about this on the Milk Road Stocks show yesterday, and I want to expand on it. A powered shell provider (PSP) owns the land, puts up the building and brings the power to it. The tenant moves in with its own chips. That split decides who carries which risk. A neocloud borrows against GPUs and customer contracts. If what compute rents for drops, the collateral drops with it. A PSP borrows against the contract too, but it is also backed by land, a building and a grid connection. Those buildings depreciate over 25 years or more. Whatever the AI trade does next year, they are still standing and still plugged in. Galaxy's Helios lease with CoreWeave shows what that looks like: 15 years, rent rising at least 3% a year, 14% yield on costs and the tenant pays for power, taxes, insurance and maintenance. The market isn't pricing that difference. When the neocloud credit scare hit this month, PSPs got sold right alongside the companies they rent to. JP Morgan said it rarely sees stocks as mispriced as this group. The other side is fair. Galaxy has one tenant at Helios, and it's CoreWeave, the name at the center of those credit fears. If CoreWeave stops paying, the building is still there, but the rent stops until someone else signs (might miss 1-2 quarters of revenue at worst?). That's why the second tenant matters more than anything else. A tenant with an institutional-grade credit rating on the 830 MW at Helios II would make it very hard to keep trading Galaxy like a leveraged neocloud bet. I own $GLXY, and I added more after yesterday's call. See my full portfolio here: link.milkroad.com/Martin

Added to my APP position. The reason is simple: I believe AppLovin can capture a single-digit share of the e-commerce market: AppLovin's CEO explains here why your next $100K on Meta is almost always your worst-performing dollar. Adam Foroughi has been buying media for twenty years, and his rule is simple: the more budget you push into one platform, the lower your ROAS (return on ad spend, the scoreboard every brand lives by). His example: a brand spends $1M a month on Meta. The next $100K on Meta does much worse than the average dollar already sitting there. Why? The cheapest, most likely buyers get reached first. Every extra dollar has to go find someone a bit less interested, at a higher price. Move that $100K to a second channel instead, and he says $1M on one plus $100K on another almost certainly beats $1.1M on one. That's AppLovin's whole e-commerce pitch in under a minute. AppLovin just expanded to e-commerce and is asking for the next dollar, the one that already works worst on Meta. There are a lot of those dollars. Online brands have put years of budget into Meta and Google, and AppLovin's ad engine only opened to every advertiser in June. We've already seen what it looks like when it works. In August I wrote about an operator who went from $0 to $160,000 a day on AppLovin in one month, and his cost per customer fell as he scaled. Foroughi admits the other side too: attribution is a headache. When a brand can't tell which channel drove the sale, the easy move is to pour more into the one it already knows. Meta also sits on years of shopper data AppLovin doesn't have. But there is still meaningful upside left to unlock, especially since AppLovin has only recently entered the e-commerce game and is still early in that buildout. I am expecting a strong Q3, with the spotlight on proving the engine still delivers and that e-commerce is gaining real traction. What are your expectations? We'll figure it out next Wednesday. Follow me @m0xt_ for more updates.

Crypto is now only 8% of Robinhood's total revenue. In Q4'24 it was 35%. So I keep wondering why the market still treats Robinhood like a crypto stock. It really isn't one anymore. At least not in the way most people think about crypto stocks. At first glance Q3 crypto looks fine: $120M, up 20% from Q2. But $26M of that came from Robinhood Chain, and chain revenue isn't tied to crypto trading volumes. It's tied to activity on the chain, which is a different thing. Strip it out and crypto trading revenue was $94M, down 6%. That's about 6.6% of total revenue. Hard to see why HOOD should keep mirroring crypto prices on that. Now the chain itself. $26M looks like a lot for something that barely existed before Q3. But a big part of it was a short-term fee spike driven by Pons, a memecoin launchpad. It pulled in a wave of traders, they launched their own memecoins, volumes exploded and chain fees went crazy. I wrote in September that this wasn't sustainable. It wasn't. On its best day in September, the chain made about $6M in fees. Now it makes about $40K a day. That's roughly a $15M annual pace, before anything is shared. But I still think the direction is what makes it interesting. Robinhood wants to depend less on speculation and market-sensitive activity, which is basically trading. A chain gives it a new kind of revenue: infrastructure. It can be home to lending, staking, trading, liquidity providing, token issuance and a lot more stuff people actually love and use. That's compelling, and worth pursuing in my view. Because the argument I just made about crypto applies to the rest of the business too. Transaction revenue was a record $867M in Q3, still 61% of total revenue. Some lines are less cyclical than others, but overall it still moves with the markets. I think Robinhood really wants to be a serious banking player, not just a trading engine. And it's still growing. Just slower. Revenue was up 12% in Q3, the weakest pace in over a year. Adjusted EPS was $0.60, flat vs $0.61 a year ago. Not a great sign. So what's the next big unlock? Robinhood's main market is the US, and it's getting penetrated fast. About 1 in 10 American adults already has a funded account, and customer growth is down to 7%. And the competition is coming. Nubank launched in the US in September. Revolut just got conditional approval for a US bank, plans to launch in 2027, and its CEO says he wants to "dominate" the US. That's why I think HOOD is getting repriced. Less like a startup that grows at any cost, more like a compounder that still depends a lot on trading activity. What is your view on Robinhood?

Galaxy posted $0.83 adjusted EPS vs. a $0.22 consensus. Nearly 4x. Nice headline. But it's not why I bought more yesterday after the call. Look at the slide: $569M of adjusted gross profit, $323M of net income, $455M of adjusted EBITDA. Big numbers, but most of it came from one place, the investment portfolio. It alone made $395M of that EBITDA. Galaxy is really three companies under one ticker: a crypto financial firm, a data center developer, and that investment book. Strip out the investment portfolio and the two operating businesses made $60M of adjusted EBITDA, up from about $1M in Q2. Their adjusted gross profit hit $150M, up from $86M. That’s the part I care about most. It’s less exposed to price volatility and market cycles. But let me actually share my comments on everything they just shared.

Earlier today, I asked where the AI payoff is outside the Mag 7. One guess: companies are piling AI on top of their existing teams, so headcount doesn't shrink and margins don't move. Microsoft just showed what the other path looks like. It offered voluntary buyouts to ~60,000 employees, about 27% of its workforce. The biggest buyout offer in its history. That’s exactly the kind of news I’m hoping for, because it’s what keeps AI trade moving forward. Microsoft's margins are already growing. But it has also put AI tools in the hands of its engineers, sellers and supply chain teams, it says it's seeing results, and now it's acting on them. That's the playbook the rest of the S&P 500 needs. Most companies aren't getting new revenue from AI yet, so the payoff has to come from costs. The same work done by fewer people, wider margins, and AI spending that finally pays for itself. The other side: 60,000 is how many people are eligible. How many actually leave is a different number, and Microsoft expects a much smaller share to take it than the ~30% who took its retirement offer this spring. Part of this is also Microsoft freeing up cash for a capex bill above $50B a quarter. But this is the third step this year. A retirement program in April, 4,800 layoffs in July, and now an offer that goes all the way up to VP level. You don't do that three times if AI isn't changing how much work each person can get done. Employees make the call in early December, then punch out on Dec 31. The real question is when the non-Big Tech crowd starts copying the playbook, or, more pointedly, how soon competitors, investors, or board members force the companies to prove AI is paying for itself. Many people are bullish on AI adoption, yet they don't expect any major layoffs. Here’s what this year has actually shown us so far: 1. UPS: 30,000 Employees 2. Oracle: 30,000 Employees 3. Amazon: 16,000 Employees 4. Omnicom: 15,000 Employees 5. Meta: 8,000 Employees 6. Microsoft: 4,800 Employees 7. Dow Chemical: 4,500 Employees 8. Cisco: Fewer Than 4,000 Employees 9. American Express: 3,840 Employees 10. Uber: 3,300 Employees 11. Renault: Up to 2,400 Employees 12. Nike: 1,400 Employees 13. General Motors: 1,300 Employees 14. Disney: 1,000 Employees 15. Citigroup: 1,000 Employees So far in 2026, the tech industry has logged about 135,000 layoffs across roughly 330 companies. That is not huge in the big picture, but it is hard to believe the pace will just stop on its own. Big layoff waves also send a message to markets. Markets can read them as proof that AI can deliver returns, which investors want to hear. This is ultimately the most bullish signal you can get. But there is a downside. If job losses keep piling up, unemployment will likely rise, which could push public spending higher at a time when many governments are already running large deficits. Will governments and regulators really sit back and do nothing to slow this down? I don’t think so. Follow me @m0xt_ for more market updates.

AI run-rate is about $200B a year today. The AI buildout through this year will need roughly $840B to $1T a year in revenue to earn a decent return. That's a 4-5x gap. The math is simple. Take how much AI capacity will be installed worldwide by the end of each year, and assume every gigawatt needs $20-25B in yearly revenue to pay off. And the bar keeps rising. By 2028, it's $2.5-3.1T a year. Everyone points to how hard it is to rent a GPU right now. But the spot market, where that shortage shows up, is only <10% of all compute. Most of it sits under long contracts. So my question is: who pays the bill? Someone has to spend close to $1T a year on AI just to justify what's being built this year. That's 4-5x what the whole world spends on AI today. Hyperscalers and neoclouds can afford the GPUs, but they're not the end buyer. They need customers renting that compute from them every year. Consumers would have to cut spending somewhere else. Companies would carry higher costs, unless AI actually makes them more money, and that takes time. Outside the M7, we’re not seeing any meaningful profit‑margin expansion yet, and that’s a concern. AI adoption will only scale when it delivers clear ROI, and we’re not seeing that today. Here’s how I see the plausible options: - Are we stuck in a limbo phase where firms are layering AI on top of existing teams so headcount doesn’t shrink and margins don’t visibly widen yet? - Are businesses struggling to operationalize AI in a way that actually produces measurable ROI? - Or are they leaning on pricey frontier models that currently swallow the gains, expecting that later optimization with cheaper models, better workflows, and tighter tooling will unlock real ROI? It’s probably some mix of all three, but at the end of the day, shareholders and the board will want to know: Where’s the AI payoff?

Galaxy reports today, and the numbers are only half the story. What management says on the call at 13:30 CET, and the guidance they give for what comes next, matter just as much. I hold $GLXY, and it's been a rough two months. The stock is down ~43% since early September, mostly because it trades with CoreWeave. The only people who can change that today are Novogratz and his team. Look at Galaxy's own slide. The black bars are what's signed: 800 MW. The grey bars are everything nobody has signed yet, and by 2030 that's 4,930 MW. That's where the upside is. So here's what I want to hear: 1. Who's taking Helios II (830 MW), and when. Last time we got "constructive" conversations and "patience is a virtue." I'd love more than that, or at least a time window. Bonus points if the tenant is investment grade or has a guarantor behind it. 2. Something about execution and CoreWeave. Is the rent coming in on time? Is Phase II still on track for Q2 2027? And what happens to Galaxy if CoreWeave gets into trouble? "We don't comment on our customer" won't cut it. 3. No new shares. In August the CFO said they don't need equity to finish the 800 MW for CoreWeave. Investors need to hear that exact line again, especially with the stock at $15. 4. Buybacks. ~$109M was left in August and the program ends around February. At 52-week lows, I'd like to hear they've been buying, and maybe that they're topping it up. 5. How they pay for what's next. Phase III still needs its final debt deal, so when does it close and at what rate? And Helios II is a much bigger build. Once a tenant signs, how do they fund it without hitting shareholders? Now let’s be honest about one thing: a good call will not fix CoreWeave’s balance sheet. And unfortunately, management burned some credibility earlier this year too. Novogratz said they would announce the next tenant by the end of summer. They didn't. Then there is Caspian. In August, they said it was “very well positioned” for base load. Five weeks later, it got Studied Load. That’s exactly why investors are more nervous this time and expect clearer, more transparent communication than we’ve seen so far this year. The numbers should be fine imo. The call decides whether $GLXY keeps trading like a CoreWeave proxy or starts trading like a landlord with highly predictable cash flows and an accelerating crypto-infra provider. One more wish, for Galaxy and everyone else. SpaceX set the bar with its earnings presentation: proper visuals and videos instead of a voice reading numbers into a mic. It's 2026. Audio-only calls shouldn't be the standard anymore. I hope more companies start treating earnings day as a chance to show investors what they're building. A few minutes of Helios on video would land better than an empty screen. I will share the update after the call. Make sure to follow me @m0xt_.

Galaxy reports tomorrow before the open. Since early September, the stock has cratered 43%, landing near $15 today, and it has nothing to do with Galaxy’s execution. It is fallout from its tenant, CoreWeave, and the broader AI rout. Over the last 90 days $GLXY has moved with CoreWeave, not bitcoin. And the last Q2 report didn't help either: an $85M net loss and a 14% drop on the day. But Q3 can be a real game-changer. Here's what I'm expecting tomorrow:

Instacart reportedly offered itself to Uber. Uber reportedly said no. As an Uber holder, that news didn't surprise me. On Thursday, it came out that Instacart, worth about $12B, is exploring a sale and had early talks with Uber and DoorDash. The stock jumped 9% to $56.44. Uber and DoorDash both fell about 3.5%. Then after the close it was reported that Uber had already walked away, and an Uber official said there are no talks right now. That leaves DoorDash as the only named buyer. Here's why I think passing was the right call. Uber says openly that it measures every acquisition against their own hurdle rate, including buying back its own stock. So a deal only happens if the price is good enough to beat that. If Uber said no, my guess is the price was too high for the upside it saw. At the right price, I think Uber would still have done it. Uber also already has a lot on its plate. It's buying Delivery Hero for $14.8B and ezCater for $2.3B, both in cash. A typical 30-40% premium would put Instacart at roughly $15.6-16.7B on top of that. Together, all three would come to about $33B, more than three years of Uber's free cash flow. Buybacks already dropped to $510M last quarter from $1.4B a year earlier. And Uber is already beating Instacart in grocery without buying it. Uber's grocery business is growing about 40% a year, Instacart's about 13%. Why pay a premium for the rival you're taking share from? The bull case for a deal is real though. Grocery is the biggest delivery category left. Instacart brings 2,200 retailers, an ads business doing around $300M a quarter, and about $1.3B of EBITDA expected this year. It would have closed the grocery gap in one move. The risk now is DoorDash. If it buys Instacart, it leads US restaurant delivery and grocery delivery at the same time. That's a much harder antitrust case, but if it clears, Uber Eats is a clear No. 2 in the US. Uber is priced at about 13x free cash flow, whereas DoorDash commands closer to 37x. To me, that gap reflects an opportunity. Uber has the scale and breadth to steadily siphon share from DoorDash over time. It is global, it is not a one-trick delivery business, and it can lean into platform network effects. That could mean outspending on customer acquisition or rolling out differentiated perks and functionality that pure-play delivery apps cannot easily match. UBER is one of the main stocks in my Milk Road portfolio. You can see it here: link.milkroad.com/Martin

Morgan Stanley's worst case for Galaxy is $16. The stock closed Friday at $14.68. TeraWulf is even further off. The bank's bear case is $17.50, and it closed at $11.32. I own both. Morgan Stanley reiterated Overweight last week, with base targets of $65 on TeraWulf and $43 on Galaxy, and called the selloff a buying opportunity. Four of the seven names in its table (TeraWulf, Galaxy, Hut 8 and Riot) now trade below the bank's own bear case. So either the bear case is too generous, or the market is pricing something worse than a bad year. I think it's pricing the second one: money for building drying up. Elon said it bluntly on Sunday. Debt markets have "almost entirely shut off access to money for any company that doesn't have substantial revenue independent of AI data centers." These landlords build mostly with borrowed money, so that sentence covers them too. He's right about the debt markets. But I think the risk is smaller here than for the neoclouds. These companies raise money to buy land, put up buildings, and connect them to power. They keep owning the land and buildings, and buildings depreciate over 25 years or more. Whatever happens to the AI trade, those assets don't go away. The rent they build against usually comes from hyperscalers or neoclouds, so the tenant risk is real. But I find it hard to imagine we'll use less of those services in the future than we do today. Whoever owns the land, buildings, and grid connection owns a valuable asset. Still, this capital-intensive phase is touching everyone, and anyone looking for funding will likely have to pay a steeper price for it, assuming they can secure any at all. But here is the other side of this trade: Morgan Stanley's TeraWulf target is about double the Street average of around $33. It also leads TeraWulf's credit line, which just got upsized to about $400M. They might be biased. And Galaxy's only Helios tenant is CoreWeave, whose funding gap banks put anywhere from $23B to $76B. The next tenant will definitely help. For all the noise, the market is pricing these PSPs at levels, and with implied upside, that are hard to look past. This is not a free lunch, but the asset base puts a floor under the downside, while the ceiling is essentially set by how much pricing power they can extract at the table. Galaxy reports Q3 tomorrow before the open. I want to see a profitable crypto business and any news on a second Helios tenant. PSPs are 13.6% of my portfolio on Milk Road. You can see the rest here: link.milkroad.com/Martin

SK hynix is reportedly weighing a special dividend of about 16T won (~$12B). If it happens, SK Square holders would get a bigger slice than direct SK Hynix shareholders. The payout would be about 23,000 won per share, roughly 1.8% of SK hynix's market value. Nice, not life-changing. SK Square owns 20.68% of SK hynix, so it would collect about 3.31T won (~$2.5B). But SK Square trades at a 46.6% discount to its holdings. So that same cash equals about 3.2% of SK Square's market value, close to double what a direct holder gets. For scale, SK Square's whole 2026 shareholder return plan is 310B won. This one payment would be about 10x that. The money would come from SK Hynix's old Kioxia stake, roughly 80% of an estimated 20T won gain. Nothing is decided yet. SK Hynix says it's reviewing options; nothing is final, and the board will decide on Wednesday with Q3 results. Back in August I wrote that special dividends could land on top of the record buyback, with the full plan coming at Q3. That's this week. The bear case is that the market won't pay for peak-cycle earnings, and a big one-off payout is what companies do at the top. Goldman just cut its KOSPI target from 12,000 to 9,000, though it stayed overweight, and says memory stocks are priced for a ~65% drop in 2027 earnings. Goldman calls that too pessimistic. I agree. In a flat-stock scenario, SK hynix trades under 2x forward earnings by the end of 2027. The catch for SK Square is that the cash lands at the holding company first. Shareholders only see it if SK Square passes it on or buys back stock, and so far its return plan says nothing about special dividends. I think this is about to change. Wednesday's SK Hynix board decision, alongside its buyback program, will be a key thing to watch. The second is whether SK Square says what it does with the money. I’m still firmly bullish on memory. It’s the largest slice of my Milk Road portfolio. See the full breakdown here: link.milkroad.com/Martin

Companies keep using more AI every month, but for two months now, they haven't paid more for it. Spending on AI tokens (what businesses pay the labs to run their models) grew just 1% in August and 0.2% in September, the slowest since last November. And it doesn't look like a summer lull. The price per token is down about 41% from the March peak, so companies run a lot more through the models while the bill barely moves. No surprise that once AI got cheaper, people started relying on it far more; that is the predictable twist known as the Jevons paradox. What bugs me is who's paying. Apollo found that 80% of the top 1% of AI spenders are VC-backed. So the biggest token buyers are mostly paying with investors' money, not profits. The bull side says this is fine. A cheap token burns roughly the same compute as an expensive one, so more usage on a flat bill is great news for chips and memory. I mostly agree. Volume is what memory sells into.. But volume only keeps growing if someone keeps funding it. If VC money slows, flat spend turns into falling spend, and the first to feel it are the neoclouds renting GPUs to labs and startups. The questions I keep coming back to is who is going to spend hundreds of billions on AI? Hyperscalers use most of their compute today for their own products, and EPS growth has been strong every quarter. But the rest of the market hasn't caught up yet. One caveat: the data below comes from a small sample of companies, so I'd treat the exact numbers loosely and the direction seriously. Microsoft and Google report Wednesday; Amazon on Thursday. I want to hear whether cloud growth is holding and who is actually renting all that compute. It is less about the backlog number and more about the quality of that customer, and how that contract is financed. Follow me @m0xt_ for more market updates and insights.

Tesla is sitting on its biggest pile of cash ever. The real question is what it's for. Cash and short-term investments hit a record $45.4B in Q3, just above the $44.7B high from March. Take out about $9.3B of debt, most of which isn't backed by Tesla itself, and that's roughly $36B of net cash. Free cash flow came in positive at $1.53B, but about $2.6B of that was timing: bills not paid yet and deposits coming in. Without it, free cash flow was about -$1.1B. Honestly, plus or minus a billion doesn't change the game for me. What matters more is that Tesla cut this year's capex to about $20B, down from "over $25B". That's interesting, though it's fair to note capex is guided up again for 2027 and 2028. Now here's where it gets interesting. Raising money is getting expensive, because everyone is out in the markets asking for it right now, and Tesla and SpaceX are no exception. But among the biggest companies, only Alphabet and Toyota hold more net cash than Tesla, while Apple, Microsoft, Amazon, and Meta all carry net debt at this point. On top of that, Tesla just signed $30B of new credit lines with banks that it hasn't used yet. So the money for Optimus/Cybercab production, factory buildout, and maybe a SpaceX deal (a merger committee is confirmed, no terms yet) is already lined up, rather than something it has to go out and raise. Tesla has never been a cheap multiple. It is still hovering near 218x trailing adjusted earnings, but it has a track record of turning bold forecasts into lived reality. While everyone else scrambles for fresh capital, Tesla appears to have financing largely under control, so its momentum likely will not be what breaks. Don’t expect their profits to suddenly jump over the next few quarters, but it’s hard to ignore where Tesla seems to be headed. A deal with SpaceX feels more like a matter of when, not if. If they end up together, the company could be worth tens of billions sooner than a lot of people expect. Keep investing little by little, and don’t count on getting rich overnight. That’s exactly what I’m about to pull off with my Milk Road portfolio: link.milkroad.com/Martin

Mastercard, Stripe and Coinbase have all paid billions for crypto plumbing. Galaxy owns one too. Look at the price tags. Mastercard closed its BVNK deal in August for up to $1.8B. Stripe paid $1.1B for Bridge. Coinbase paid $2.9B for Deribit, and Ripple paid $1.25B for Hidden Road. This week Visa was reported to be in talks for Fireblocks at about $2.5B, which Fireblocks denies. Big payment companies and banks keep paying $1-3B for crypto infrastructure with recurring fees and institutional clients. Now apply that to Galaxy. Leave Helios out, and Galaxy still runs a trading and lending desk with over 1,700 counterparties, plus $7.1B in assets it manages or stakes. BlackRock and Morgan Stanley both picked it as a validator for their staked ETH funds, and BNY signed it up as a design partner for staking. Last quarter, that business made about $66M in gross profit, about $17M of it from fees, staking, and custody. Here's where it gets interesting. Run that business for a full year, and it makes roughly $160-195M of gross profit from trading and lending, plus about $70M from fees, staking, and custody. Put about 5x on the trading side and 8x on the fee side, in line with the deals above, and you get roughly $0.8-1B plus about $0.55B. That's about $1.3-1.5B in total. One broker says the stock gives the crypto business "effectively no value", because everyone is staring at the data center campus. For context, they still bled $9M last quarter and crypto deal flow slowed in Q3. But earnings hit Tuesday, and I’m betting the crypto segment finally turns profitable and catches the market off guard. Galaxy is still my largest position on Milk Road PRO. See my portfolio here: link.milkroad.com/Martin