Stock & crypto market Enthusiast. ✍️ substack.com/@financialpulse… ✍️ Not financial advice. Always DYOR. $DUOL $NOW $ORCL $MSTR $GRAB & more

Between ORDER and CHAOS
I saw markets yesterday get especially bearish on $ORCL regarding new escalations with buildout delays in the New Mexico, and while those kind of things are to be expected especially with buildouts of that scale where politics get in a way, I like to use these times to reflect on the positives about the company and once more it takes me back to their recent earnings which demonstrated strength in multiple directions.
Just finished with the $ORCL earnings call and more goodies if you ask me. First, the more obvious one is revenue acceleration, and it's one thing to say it, but in Oracle's case, they have strong support behind it: Q4 24: 14.3B ➡️ Q1 25: 13.3B ⬇️ (seasonal decline) Q4 25: 15.9B ➡️ Q1 26: 14.9B ⬇️ (seasonal decline) Q4 26: 19.2B ➡️ Q1 26: 19.3B ⬆️ (rise). This is something I am hoping to see in the $PATH case as proof of acceleration but still waiting. Now back to the topic. The majority of the newly secured contracts came in the form of some kind of prepayment or bring-your-own hardware (26 billion). Half of their RPO is expected to be converted into revenue in the next 36 months. Total RPO: 664B During the next 3 years: 332B Another positive was that, yes, Capex was 28 billion, but it was a mix of customer prepayments and Oracle's own cash (it was not financed completely from Oracle's side). The next piece of information answered two of my questions (discussed in more detail on my Substack): "Of all the GPUs that came up for renewal in Q1, that capacity was renewed or resold at a 20% premium to prior contracts. The majority of those GPUs are 4 years or older. " First question is: can capacity remain highly utilized? 97.9% of total utilization answers that. The second question is: how long are GPUs economically useful? The fact that GPUs can be resold at a 20% premium when they are 4+ years old gives another positive affirmation. Lastly, an interesting piece of information they shared is that when data centers go live, for example, if you have a big capacity like 1 GW, not all of it goes active at once; it goes in multiple stages, in smaller batches. Price action managed to turn negative on the day, but I am not worried. I saw all I wanted to see in this earnings, and September usually is not kind to stocks, and the hot print in PPI and CPI certainly didn't help.
3
172
I’ve mentioned before that $GRAB Financial Services segment is, in my view, its most interesting growth vertical, but also potentially its biggest source of risk. That’s why I’m watching a few key metrics that could provide early warning signs if credit quality starts deteriorating. One of the most important is non-performing loans (NPLs). Management usually says credit quality remains within expected levels, but we don’t get a clean reported NPL ratio. So I went through the annual report to build a reasonable proxy. Grab classifies loans that are 91–120 days past due and more than 121 days past due as credit-impaired. For 2025: ♦️ 91–120 days past due: $9M ♦️ >121 days past due: $11M ♦️Total credit-impaired loans: $20M ♦️ Gross loan portfolio: $1.278B That gives: $20M / $1.278B = ~1.56% Using the same methodology for 2024 gives ~1.88%. This is NOT an officially reported NPL ratio, but I think it’s a useful proxy because it is based directly on the portion of Grab’s loan book that the company itself classifies as credit-impaired. And the direction is encouraging: Credit-impaired loan ratio: 2024: ~1.88% 2025: ~1.56% …while the gross loan portfolio more than doubled. For now, loan growth is clearly not being accompanied by an explosion in credit-impaired loans. That’s exactly the relationship I’ll keep watching as Financial Services becomes a larger part of the $GRAB story.
2
139
$GRAB CEO Anthony Tan just bought 10.35M shares at ~$2.89, roughly a $30M open-market purchase. Interesting timing after the recent sell-off. I’ve just published recently my deep dive into Grab: its growth story, financial services expansion, Atome acquisition, risks, and valuation. Article is on Substack (link in my bio). Usually my timing is not perfect but thos is as close to perfect as it gets. 🤞😅
4
152
You want to tell me I finally decided to start a position in $GRAB and CEO joins the party, well let's frikin go to the moon (6$ will do it for now 😅🤞).
$GRAB CEO Anthony Tan buys 10,350,000 shares of Grab at $2.89/share. Anthony has been a net seller on the open market since the company went public. This is first open market purchase. Obviously, for him to buy $30M of stock while his net worth is around $300-400M shows how significant of a purchase this is and how much he thinks the market has lost their minds on this name. We are now looking at a public valuation LOWER than what Grab was worth in the private markets. Their enterprise value is $7B and they will profitably do $4.2B of revenue this year. Really nice to see an insider buy — it has been a tough hold but the broader thesis, in my opinion, is being executed in a VERY strong way with a horrible macro backdrop . For a CEO who has been criticized for many insider sells (many of which were pre-planned and also deserved given the guy founded the company and wasn’t liquid for a decade) it is nice to see him put almost 10% of his net worth into the company he started 15 years ago via purchases on the open market. Grab Holdings President & COO Alexander Charles Hungate also bought $867K of $GRAB with 299,571 shares at a $2.89 avg. The markets are the greatest game on earth, they will test your patience, they will turn your stomach inside out, they will drive you crazy as the price of an asset goes down while fundamentals get better, every bear case will make more sense as you see the price go down, but one thing is clear: execution and compounding earnings wins in the end. It’s nice to see a C-suite that believes in that themselves and takes advantage of that with a sizeable amount of their personal networth. The story continues!
1
15
351
I like what this could mean for my $MSTR bag.
My official bull market indicator has flashed. Price just broke above the 2-day 200 MA cloud. Bitcoin is back in a bullish regime, unless it falls below. Each time price reclaimed the cloud after a retest on the 200-week MA, price rallied hard. Embrace nuance, risk takers.
4
157
What a start of the week. $BTC ripped 3k. $SOL is holding strong. We are cruising through September pretty decently for now, fingers 🤞 it continues.
3
89
It's interesting how $MARA was performing well yesterday on the fuel of data centers (still waiting to secure it's first tenant though), and today it's rebounding even more on the bitcoin:native strength.
2
120
One interesting thing I found while reviewing $GRAB financials is that the loan book is growing much faster than the loss reserve, which may be a positive sign if asset quality holds up. From the numbers I reconstructed across company reports: ♦️ Q1 2025 gross loan portfolio: $625M ♦️ Q2 2025: $781M ♦️ Q4 2025: $1.18B ♦️ Q1 2026: $1.40B ♦️ Q2 2026: $2.30B At the same time, the implied credit loss reserve/allowance moved from: 💠 $59M in Q1 2025 💠 $73M in Q2 2025 💠 $98M in Q4 2025 💠 $117M in Q1 2026 💠$153M in Q2 2026 That means reserve coverage versus the loan portfolio went from roughly 9.4% in Q1 2025 and 9.35% in Q2 2025 down to about 6.65% by Q2 2026. So while the nominal reserve is rising, it is rising much more slowly than the loan book itself. That can suggest one of two things: 1. underwriting and portfolio quality are holding up well as the book scales 2. the mix is shifting toward lower-risk products/borrowers One important note: these are not quarterly P&L credit loss provision expenses, but rather the balance-sheet loss allowance/reserve implied by Grab’s disclosures. Also, comparing these periods takes some work because Grab changed how it presented the loan portfolio (that's why Q3 25 is still missing): ➡️ older reports showed net loan portfolio. ➡️ later reports started showing gross loan portfolio. ➡️ for newer quarters, the reserve can be reconstructed by subtracting net loan receivables from the reported gross loan portfolio. This is one of the financial services metrics I would keep watching closely going forward, especially together with NPL trends, delinquencies, and impairment expense.
1
2
175
Well, I didn't hope for this kind of bitcoin:native performance to close the week. Bitcoin is closing up to 81K, up ~6% for the day. Strategy is closing up to $150, up ~11% for the day. Feels good to see $MSTR running the portfolio for a change.
2
179
So data center compute plays were up today nicely. My current picks $ORCL and $MARA both posted +5% daily gain. $AGPU was an exception with +18% in a day. My personal take from usage of Work/Computer Use of Chat Gpt for work (Excel / Power BI mostly) is that I am burning tokens faster than eating sweets (and man I like sweets). I almost burned my whole weekly usage 3 days before it completely restart. Based on this we are not nearly enough bullish on the data center plays, however much you think is enough, double it, no triple it.. you get the point.
2
8
531
One thing I really liked at $ORCL last earnings is that despite acceleration in their backlog conversation they still manage to increase overall backlog from 638 Billion to 664 Billion with majority of the new contracts coming in a form of a prepayments. Obviously as time passes less and less focus will be on the size of the backlog and more eyes on the speed of the conversion to the actual revenue.
2
3
171
So the Fed decided to raise rates by 25 bps. The decision was unanimous. 16 of 18 members see at least one more hike by the end of the year.
The market may be overestimating the probability of a Fed hike tomorrow. A hike is still base case. But when the market gets this confident, I think it is worth looking at the counterarguments. 1. This does not look like 2022. Before the first hike in March 2022, the Fed spent months preparing markets for liftoff. The communication path was much clearer. This time, expectations shifted extremely fast. As recently as September 9, a Reuters poll still had the majority of economists expecting the Fed to hold. Now, 85% expect a 25 bps hike. That is a huge repricing in less than a week. 2. One hot inflation print does not necessarily force an immediate response. Inflation is clearly a problem again, especially with oil above $100, but the Fed could still choose to wait for more confirmation rather than react immediately to what may partly be an energy-driven shock. Hold does not have to mean dovish. They could hold tomorrow and Warsh could use the press conference to deliver an extremely hawkish message, keeping a hike firmly on the table for the next meeting. 3. The political backdrop is unusual. Trump continues to publicly push for lower rates, while the U.S. is heading into the midterms. That should not determine Fed policy, but it is part of the environment in which this decision is being made. None of this means the Fed will hold. The inflation data, oil shock and recent economic strength all make a hike completely defensible, and it remains the most likely outcome. My point is simply this: What happens if we get a hold decision tommorow?
1
189
I initiated a new position in a company called $GRAB. I still need to write my full deep-dive breakdown, but in short, I like them as a main player in Southeast Asia as a super app for ride-hailing, delivery services, and, most recently, a serious expansion into financial services. Strong growth is still ahead, with a recent upgrade to 30% revenue CAGR in the period 2025 - 2028. With the price at multi-year lows, they became an interesting diversification play for me from my heavy AI-related portfolio.
5
167
The market may be overestimating the probability of a Fed hike tomorrow. A hike is still base case. But when the market gets this confident, I think it is worth looking at the counterarguments. 1. This does not look like 2022. Before the first hike in March 2022, the Fed spent months preparing markets for liftoff. The communication path was much clearer. This time, expectations shifted extremely fast. As recently as September 9, a Reuters poll still had the majority of economists expecting the Fed to hold. Now, 85% expect a 25 bps hike. That is a huge repricing in less than a week. 2. One hot inflation print does not necessarily force an immediate response. Inflation is clearly a problem again, especially with oil above $100, but the Fed could still choose to wait for more confirmation rather than react immediately to what may partly be an energy-driven shock. Hold does not have to mean dovish. They could hold tomorrow and Warsh could use the press conference to deliver an extremely hawkish message, keeping a hike firmly on the table for the next meeting. 3. The political backdrop is unusual. Trump continues to publicly push for lower rates, while the U.S. is heading into the midterms. That should not determine Fed policy, but it is part of the environment in which this decision is being made. None of this means the Fed will hold. The inflation data, oil shock and recent economic strength all make a hike completely defensible, and it remains the most likely outcome. My point is simply this: What happens if we get a hold decision tommorow?
1
337
Dayum 90% of the rate hike on the next meeting.
1
63
While most of the AI sector is getting rekt today, I want to focus on one positive: companies that are already showing how AI can make their existing businesses materially better. Duolingo is one of the clearest examples. It took the company roughly 12 years to build its first 100 courses. Then, using generative AI, shared content systems, and internal automation, $DUOL launched 148 new language courses in less than one year. And the acceleration is continuing. In Q1 2026 alone, Duolingo published 20,500 course skills, compared with: ♦️ 7,100 per quarter in 2025 ♦️ 1,800 per quarter in 2024 That is more than a 10x increase in content production velocity in just two years. This is what I find interesting about AI. The value is not only in companies selling GPUs, models, or infrastructure. It is also in businesses that can use AI to reduce the time and cost required to expand their product, experiment faster, and ultimately serve more users. Duolingo finished 2025 with 52.7M DAUs, and management believes it can reach 100M daily active users in the medium term. AI alone will not get them there. Product quality, retention, and organic growth still matter. But if AI allows Duolingo to build more content, expand into more languages and subjects, and run product experiments dramatically faster, it becomes a very real tool for reaching that target. That, to me, is one of the more interesting parts of the AI story: not just who sells AI, but who becomes a better business because of it.
1
76
Just finished with the $ORCL earnings call and more goodies if you ask me. First, the more obvious one is revenue acceleration, and it's one thing to say it, but in Oracle's case, they have strong support behind it: Q4 24: 14.3B ➡️ Q1 25: 13.3B ⬇️ (seasonal decline) Q4 25: 15.9B ➡️ Q1 26: 14.9B ⬇️ (seasonal decline) Q4 26: 19.2B ➡️ Q1 26: 19.3B ⬆️ (rise). This is something I am hoping to see in the $PATH case as proof of acceleration but still waiting. Now back to the topic. The majority of the newly secured contracts came in the form of some kind of prepayment or bring-your-own hardware (26 billion). Half of their RPO is expected to be converted into revenue in the next 36 months. Total RPO: 664B During the next 3 years: 332B Another positive was that, yes, Capex was 28 billion, but it was a mix of customer prepayments and Oracle's own cash (it was not financed completely from Oracle's side). The next piece of information answered two of my questions (discussed in more detail on my Substack): "Of all the GPUs that came up for renewal in Q1, that capacity was renewed or resold at a 20% premium to prior contracts. The majority of those GPUs are 4 years or older. " First question is: can capacity remain highly utilized? 97.9% of total utilization answers that. The second question is: how long are GPUs economically useful? The fact that GPUs can be resold at a 20% premium when they are 4+ years old gives another positive affirmation. Lastly, an interesting piece of information they shared is that when data centers go live, for example, if you have a big capacity like 1 GW, not all of it goes active at once; it goes in multiple stages, in smaller batches. Price action managed to turn negative on the day, but I am not worried. I saw all I wanted to see in this earnings, and September usually is not kind to stocks, and the hot print in PPI and CPI certainly didn't help.
1
4
367
Diving into the numbers and I can say I like what I see. RPO increased from 638 to +664 billion. Q1 total revenue grew 30% to 19.3 billion (slightly beating their own range 27-29% expected). Cloud Infrastructure up 121%, to 7.4 billion, and this comes after in Q4 26' they grew 93%! Operating cash flow of 23 billion. Delivered 800 MW of additional capacity, and almost tripled the capacity delivered in Q4 26'. Regarding guidance they reaffirmed they FY revenue of at least +90 billion (+34% YoY growth). Also, they slightly bumped their non GAAP EPS from $8.05 to $8.1 Now, to be clear this growth is having it's price. Despite record operating cash flow in Q1 27', free cash flow came at negative 5 billion. They completed their 20 billion ATM offering during Q1. So the growth continues to be funded via. debt and dillution of the shareholders but this is happening on the back of their insane demand which still manages to grow. Stock price is up ~10$ after the report, erasing the dip we got during the day mostly thanks to the weakness in the general markets (also tommorow we have CPI print). I saw the what I needed to see which is growth remains robust. Tommorow I will digest earnings call and went into more details of $ORCL earnings and I will give more of my insights.
3
82
ManI hope I am wrong but I have a feeling $ORCL will get rekt tonight. 😭 The market seems ready to punish anything. It's my biggest position btw so as long as I like what I see I am fine, price action is short term noise.
2
189
$ORCL reports its earnings today. What will the market value more today: a growth story or a capex story?
2
100