On nous vole des sommes folles pour les reverser à la terre entiÚre, mais lorsqu'il s'agit de restaurer notre patrimoine, ça passe par des dons volontaires.
@Tino_Bourse@JulienTechInvst Bonjour Tino, PRU a 8 USD, ton avis sur lâextension possible pour cette nouvelle jambe de hausse. Je vise 24-25 USD possible.
$CELH Q4â25 EARNINGS HIGHLIGHTS
đč Revenue: $721.6M (Est. $639.14M) đą; UP 117% YoY
đč Adj. EPS: $0.26 (Est. $0.19) đą
đč Gross margin: 47.4%; DOWN 280 bps YoY
đč Adj. EBITDA: $134.1M; UP 113% YoY
đč FY25 revenue: $2,515.3M; UP 86% YoY
Segment Performance:
đč North America revenue: $699.5M; UP 124% YoY
đč International revenue: $22.1M; UP 9% YoY
Other Metrics:
đč Portfolio share of zero-sugar U.S. energy category growth (2025): 33% (of $3.3B growth)
Capital Return:
đč Debt repayment: $197.8M
đč Share repurchases: $39.8M
Commentary:
đž â2025 was a defining year for Celsius Holdings as we delivered record full-year revenue of $2.5 billion...â
đž â...we reached an approximate 20% dollar share of the U.S. energy drink category in Q4 2025.â
Thanks for all the support on my $UPST interview with Paul. I spent a lot of time thinking about the questions I wanted to ask him, so I want to unpack some of my thinking.
For the first part of the Q&A, I wanted to dive into the significance of Paul's experience in building the models from Zero to One, and why that's so important as he steps into the role of CEO.
From there, I wanted to explore just how far the technology and business has progressed since 2022. While Upstart is exposed to all the noise of fintech, at their core, Upstart is fundamentally an AI company sitting at the application layer. Model improvements are at the core of the thesis.
Having said that, it is shocking to me that analysts do not spend more time seeking a greater understanding of the models. But if we take it a layer deeper than that, it's easy to say that Upstart's technology has improved since 2022, but for those not looking deeply it's hard to quantify.
Paul gave very thorough answers here and even quantified the improvements in tech, stating that UMI would have allowed Upstart to avoid 80% of the 2022 disruption to credit outcomes. Similar language was shard at AI Day.
I then moved onto the future, where Paul elaborates on how all of these developments impact Upstart's long-term prospects. He also provided an in-depth explanation on some of the asymmetry he is seeing in Upstart stock, evidenced by his $3.9M insider buy. That is an extremely large insider purchase.
With all of that now as context (his experience in building the models, Upstart's turnaround & improved technology, rosy future prospects & insider buying) I then transition the conversation to the models.
Upstart is rarely asked for color on the granular details of the data flywheel or machine learning techniques on the conference call. The analysts covering the company appear to be skewed more towards financials than they are technology, with some exceptions of course.
The big takeaway here is that Upstart's data flywheel is exploding. It took them 14 years to hit 100M repayment events. It will only take them a few years to double that. And that weaves into my next question on margins for Auto & HELOC. These products, when you exclude intention margin levers (such as prioritizing life-time value/market share efforts) will grow at expanding margins as the underwriting gets better.
Put together the largest TAM possible (lending) with the most transformative technology humans have every created (AI) and not only do you have a business with extreme growth prospects, but you also have a business that can grow into that market increasingly more profitable as the technology gets better.
That is incredibly powerful. Upstart is effectively going after the lending trifecta of growth, profits, and credit performance. Traditional lenders can not achieve all three. If you grow your originations, normally you sacrifice profits/credit performance. That's because traditional underwriting is static (or constant) and not improving over time.
Upstart has dynamic underwriting. Their models are continuously and autonomously updating to real time borrower behavior and macro signals. New models have a much shorter feedback loop because Upstart utilizes PTCC (making prepayment and default predictions by month) whereas legacy models only look at the terminal state of a loan (paid or default). In my opinion, Upstart's models are not only improving, but the rate by which they are improving is also accelerating.
Dynamic underwriting allows for origination growth with better credit performance, which leads to more profits. Upstart accomplished this seemingly impossible trifecta in 2025.
Now, the fruits of that labor are currently being overshadowed by 1) the macro and 2) by intention efforts to take market share. Upstart is forgoing the maximization of take rate in exchange for a larger piece of the lending pie. That's why I then asked Paul about Upstart's marketing strategy.
His answer makes it clear Upstart is spending increased attention on this front. They are in the process of a complete rollout of a new mobile app that will deepen borrower engagement with the brand. The launch of new products also enables Upstart's marketing to cast a wider net, making marketing dollars more efficient.
And so naturally the last piece of the puzzle after we understand that 1) Paul's experience in building the models is foundational to Upstart's next chapter 2) Upstart's technology today is much better than in years past 3) Paul sees an incredibly rosy future for the business, supported by a sizable insider purchase 4) the data flywheel of models is improving and accelerating 5) Upstart is waning back margins short-term intentionally for market share ... becomes partner growth.
If Upstart is to be everywhere - they have to integrate with more partners. In particular, I really want to see them partner with a big bank. So, I asked Paul about that. The answer? Big banks 1) need more than just personal loans and 2) are naturally risk averse to new technology by nature.
Given Upstart's evolution across product set and increased positive press about AI-driven value creation, these obstacles seem surmountable. That gives me confidence that if Upstart's technology is truly the best, in the end they will win with banks.
So that brings you 360 on my thoughts for this interview. The Upstart thesis at the core is one of AI and technology, with the macro effects pouring gas or water that fire. They have grown quite substantially in the face of elevated macro risk (UMI). That is only possible if the underwriting is getting so much better that they can grow via model improvements (with lower APRs) despite elevated macro risk.
I believe Upstart is building something very special and very unique. It's been a volatile journey, but the fundamentals are improving and management is executing. Hope you enjoyed the discussion.