Is $NU a Buy After the US Expansion? 🤔 👇️ Less than a year ago, Nu revealed a very ambitious goal of becoming a major US bank. Originally, there were estimates that the company could enter the market in 2027 or even 2028. Yet, as of yesterday, their US operations are live! Furthermore, the company had its first-ever presentation from Miami, where CEO David Velez and co-founder and US CEO Cristina Junqueira presented many interesting new developments.

$SOFI is a broken stock, not a broken business! 2018 ➡️ Today: - Revenue $268M ➡️ $4.3B - EBITDA -$227M ➡️ $1.1B - Net Income -$252M ➡️ $646M - EPS -$4.02 ➡️ $0.51 - Members 652K ➡️ 15.8M - Loans $11.7B ➡️ $47.4B

Next week $MRDN Expanse Studios will participate in the Global Gaming Expo in Las Vegas. A reminder that Expanse Studios develops casino games for Meridian's own operations and for third-party operators internationally. This segment is in aggressive scaling mode, making new games with a portfolio of 100 proprietary titles and more than 1,800 casino clients worldwide. The priority for Expanse is now to get Ontario and New Jersey licenses, and this expo will help them showcase their offering before these licenses are secured. At Expo the studio will present its newest game titles, including recent slot and crash game releases, to the operators, aggregators and platform partners who attend the show. Disclosure: This is Issuer-Sponsored Research

The best reason to believe $NU can succeed in the US is in Mexico. Nu hit 16M customers there, and they're progressing faster than Brazilians did at the same stage. ARPU in Mexico reached $12.3 in Q2 2026, against $5.6 for Brazil at the same point in its journey. Every market teaches Nu something it carries into the next. It took the Brazil model, adapted it, and Mexico ramped faster. They will apply all these learnings from Brazil, Mexico, and Colombia in the US!

Tether's $USDT makes $10B profit from $180B stablecoins in circulation. Now imagine a stablecoin issued by a regulated US bank, instead of some shady company that launders money for Iran and North Korea. This is what the Sofi x Mastercard partnership is all about. This partnership allows $SOFI to move money instantly at any time, instead of waiting days for traditional banks to process transactions. Issuers and acquirers can choose to settle transactions in SoFiUSD, allowing for 24/7/365 instant settlement. Simply put, by using their own digital currency, they can avoid paying expensive fees to middleman banks and keep more of that profit for themselves. Sofi will begin by settling all its Mastercard transactions using SofiUSD, demonstrating the usefulness of this technology. Then, through its Tech Platform business, other companies will use this faster technology. There are huge applications for this technology in: - Remittances - B2B Payments - Bank-to-Bank Settlements $SOFI has the potential to make billions in fees from its stablecoin. Capturing 10% of Tether's volume would mean $1B+ in new revenue at 80%+ EBIT margins. $SOFI's entire net income last year was around $600M. In a few years, this business could be the same as the LPB is today.

$NU is down 28% from its highs, trading at a P/E of 19. Analysts expect earnings to grow 25% a year through 2028, which would double EPS and take the P/E down to 10 on today's estimates. My view is that those estimates are way too low, and the company could be trading for 6-8x 2028 EPS. In my opinion, analysts are underestimating earnings growth that will come from Mexico. Nu Mexico became break-even 2 quarters ago, and the country is scaling faster than Brazil was. Nu is implementing the lessons from scaling in Brazil in Mexico, and in the next 2 years the unit will deliver really strong growth.

This was the most important thing in $MRDN Q2. Revenue grew +16%, yet operating costs fell 8.5%. When sales climb and costs drop at the same time, almost every extra dollar of revenue drops straight to the bottom line. That's operating leverage, and it's the moment a business stops burning cash and starts compounding it. It's why operating profit swung to +$2.44M from a -$2.3M loss a year earlier, and why adjusted EBITDA grew over +40% Y/Y. $MRDN spent years building its platform, its network of 740 betting shops, and its market licenses. Now it can run far more volume through that same infrastructure without adding much cost. Disclaimer: This is Issuer-Sponsored Research

Memory demand is growing faster than the supply, yet the stocks are all down a lot, which one is the best buy? - $MU at 6 FWD P/E - $SKHY at 4 FWD P/E - $SNDK at 7 FWD P/E - Kioxia at 4 FWD P/E - Samsung at 4 FWD P/E

I think some investors are misunderstanding what SemiAnalysis ClusterMax measures. 1) It doesn't measure profitability. 2) It doesn't measure the AI opportunity. 3) It doesn't measure whether the stock is a BUY. It measures: 1) How fast providers detect dead GPUs, drain bad nodes, and swap hardware without stalling training. 2) True network efficiency (whether inter-node fabrics deliver advertised bandwidth) 3) Cloud software maturity (how well managed Kubernetes, Slurm, and storage run) Furthermore, they clearly a label that this is managed cluster, not Bare Metal. You can argue about methodology and bias, but it is not surprising to me that IREN has a bad ranking here. They literally just started doing managed clusters, while Nebius has been doing them since day one. I am an IREN bull, and other bulls shouldn't just scream "BIAS" and instead realise that IREN needs to do a lot of work to catch up to Nebius and Coreweave. Moreover, SemiAnalysis is mostly talking about the sites in Canada that are actively being converted from BTC mining to AI, of course, there will be issues in the process. The sites in Texas will look much better. Iren more than doubled the employee count in 2026, they will hire more people to find and fix various issues that will undoubtedly arise. You don't build a $100B+ 5GW AI Cloud without dealing with countless issues. So overall, Iren can still be significantly profitable and deliver great returns for shareholders, even if everything that SemiAnalysis says is true. They have great data center assets with real cost advantages. Most importantly, all capacity is sold out, as the demand is incredible.

No, $ORCL is not bankrupt and is not insolvent, as some people have claimed. $ORCL needed power for a giant AI campus and couldn't wait years for the grid. $BE got the first system running in 55 days. That's 35 days ahead of its own 90-day target. As a result, Oracle liked it so much it expanded the deal to up to 2.8 GW. However, yesterday Oracle filed a force majeure notice on that New Mexico campus, Project Jupiter, because the natural gas pipeline meant to fuel it won't be built in time, slipping from August 2026 to February 2027 because of permitting delays. Force majeure clauses excuse a party from contractual obligations when events outside its control get in the way. Oracle is using it to reserve the right to delay rent. Oracle is attempting to put off payments should the data center get derailed and fail to come online in 2028 as planned, and it intends to stay on as the site's main tenant. No fuel on site means Bloom can't switch its cells on there yet, so some of that revenue could push out a few years. Bloom says it can redirect the equipment to other sites, and both sides insist the project is still on track. Nevertheless, this is a clear signal that AI data center buildout won't be without issues. China is winning the energy race because there are no regulatory delays in China.

18 months ago the bear case on $MRDN was simple, too much merger debt. That case is falling apart: - Key debt -46% Y/Y to $34.6M - Net debt -65% to just $9.4M - Net leverage down to 0.39x - $17.3M cash in the bank A small-cap drowning in debt has no room to invest and thus was punished hard. One with almost no net debt and growing cash flow can fund its own push into Brazil, Mexico, and the US without requiring additional capital. $MRDN reduced debt while still growing revenue and turning a profit. They could essentially get rid of all the debt with less than 2 years of ADJ EBITDA. Disclaimer: This is Issuer-Sponsored Research

The Amazon of Latin America vs the real thing. Which do you buy? - $AMZN at 27 FWD P/E - $MELI at 44 FWD P/E

People overestimate how well the average US banking consumer is served. $NU has a real opportunity, as: - 5M+ American adults have no bank account - Check-cashing kiosks process around $100B a year, pulling $2B in fees from people with nowhere else to go - 10-20% of US households earn just $20-30K a year, a pool of 30-60M people paying high fees to legacy banks for poor service $NU built its scale by onboarding exactly these customers across Latin America. The company is well positioned to continue doing the same in the US.

Better digital bank to own for the next 5 years? - $NU at 14 FWD P/E - $SOFI at 23 FWD P/E

$MRDN's in-house game studio, Expanse Studios, is growing at a pace that's investors shouldn't ignore: - Revenue +138% Y/Y - Players +67% Y/Y - Pay-ins +143% Y/Y - Games now on 1,881 operator sites, up from 1,519 last quarter Expanse builds the actual casino games, slots, table games, and the fast crash games exploding across emerging markets. It launched 18 new titles last quarter, taking its library to 95. Being inside a larger group is a real competitive advantage. $MRDN's own aggregator distributes Expanse games cheaply, its own casinos give them guaranteed shelf space, and the group can see which titles make the most money across billions in wagers, then build more of them. It just signed deals with Bragg and a Flutter subsidiary, won certifications in Latvia, Colombia, Portugal, and Slovenia, and is waiting on approval in New Jersey and Ontario. Once this segment scales, it could become a meaningful business for the company. Disclaimer: This is Issuer-Sponsored Research

Some quick math on $NU's US potential. Say Nu wins 10M US customers by 2033. Its current ARPU is $17 a month, and wealthier US customers could realistically generate 5x that. That's $1B a month, roughly $12B a year. Nu's entire 2025 revenue was $11.2B. The US alone could one day match everything Nu earns today from Brazil, Mexico, and Colombia combined. This is a significant opportunity that shouldn't be so easily dismissed. You don't sponsor Messi if you don't have huge ambitions.