Start the printer, Powell.
Private credit liquidity is tightening.
Blue Owl will permanently restrict redemptions in its private retail debt fund, shifting from quarterly withdrawals to episodic capital returns as assets are sold.
As one of America’s largest private credit managers, don’t underestimate how big this is.
Investors had previously been allowed to redeem up to 5% of net assets per quarter.
Redemptions were accelerating.
$150 million was pulled in the first nine months of 2025, up 20% YoY.
Third quarter withdrawals nearly doubled to $60 million, about 6% of NAV.
The firm is now selling $1.4 billion in loans across funds, roughly 30% of the retail vehicle’s assets, to generate liquidity.
Private credit has absorbed hundreds of billions from retail investors seeking yield with limited volatility.
When investors want out, the exit gets smaller.
Yield was abundant. Liquidity was conditional.
This is how pressure builds in the credit system.
And with this credit strain, policy will likely respond.
Do you think Powell is warming up the printer now?
$ONDS FIFA Award happening this week folks?!!?
"The Program Executive Office has already begun its work and, this week, will finalize a $115 million investment in counter-drone technologies that will be critical in securing America250 and 2026 FIFA World Cup venues"
https://t.co/t7KfYVwX8p
$GRAB GrabMart
Most people likely aren’t familiar with this segment because Grab doesn’t break it out clearly in its financial reports.
However, it is one of the highest potential segments due to SEA’s inherent characteristics: 🧵👇
GrabMart was first piloted under GrabFood in Nov 2019. When COVID hit in 2020, GrabMart was perfectly positioned to capitalise on the demand. From April to Sept 2020, it grew 50% week-on-week. (You read that right!)
A key driver of this was GrabFood, with 85% of GrabMart users directly onboarded through GrabFood. GrabMart managed to scale effortlessly due to their delivery network (GrabFood) and operational footprint.
Fast forward to 2025, GrabMart is present in all 8 Grab markets across SEA and is growing faster than the overall deliveries segment. For context, the Grab Deliveries segment (which GrabMart falls under) is at a ~$13B run-rate. GrabMart is estimated to be ~10% of that, equivalent to a $1.3B annual run-rate.
It has onboarded over 50,000 convenience, pharmacy and grocery outlets, with sub-60 min delivery. But more importantly, it’s started to go full $AMZN. It acquired Jaya Grocer and Everrise Supermarket and has integrated them as micro-fulfilment hubs; serving dual roles, as retail footprint and instant-delivery nodes.
Something that few seem to realise, is that Grab is going for the entire food market.
The food market can really be split into 3 broad categories: dining out, dining in, and food delivery.
Grab already dominates the food delivery piece, with >50% market share in SEA. It is increasingly dominating the dining out piece, with the acquisition of Chope (the most popular reservation app in SEA) and HungryGoWhere (restaurant discovery site), and its new feature “Grab Dine out Discovery”. Now, Grab Mart intends to bring it home (no pun intended), with grocery deliveries for consumers who intend to cook at home.
Home-cooking is more prevalent in Southeast Asia than in almost any other region. In SEA, 7.2 meals a week are cooked at home, 15% over the global norm. Several regions across the world have also seen cooking fall back sharply post-pandemic, but SEA’s average has held flat, suggesting that the habit is resilient.
In many other markets, this number has dropped post-COVID, but in SEA it has remained resilient, thanks to two big structural drivers:
(1) multi-generational households where entire families eat at home together.
(2) the significantly lower cost of dining in versus eating out.
Despite the majority cooking at home, online grocery penetration is barely 3% of total food spend. GrabMart has a multi-billion dollar TAM to chase. The online grocery market in Southeast Asia was $26.6 billion in 2024, and is expected to grow at a 17% CAGR, hitting $112 billion by 2033.
As of FY24, Grab had just a 5% market share. I believe it gets up to a 20-30% market share by 2033 (it already dominates the adjacent verticals and is just getting started), which will equate to $22.4B GMV on the low end. GrabMart has the potential to eclipse the entire Grab on-demand GMV in 8 years.
There’s also the quick-commerce angle, which is a no-brainer adjacency to expand into. Grab is already piloting 20-30min GrabMart Express in cities like Jakarta and Manila. Quick-commerce is expected to be a $30B market by 2030. If Grab captures just 10% of this, that equates to another $3B in GMV.
And GMV is just the starting point. While groceries and deliveries are known for low margins, the unlock lies in retail media, the same way Instacart monetises its grocery base. Retail media offers structurally higher margins (40-70%) and Grab is already tapping this.
Every +1% of ad revenue as % of GMV drops 60 to 80 basis points to segment adjusted EBITDA. It’s the most scalable, high-margin part of the business.
Beyond that, GrabMart also feeds into the entire Grab engine, which is what makes it particularly exciting to me. GrabUnlimited for instance, now includes Mart benefits. This works just like Amazon Prime and Instacart+, increasing frequency, stickiness and LTV. GrabMart also supercharges the FinTech engine. Grocery purchases are frequent and predictable, which gives GrabPay and Grab’s digital banks daily transaction flow and behavioural signals.
Why is this important?
Groceries builds habit, habit builds deposits, and deposits lower the funding cost, which reinforces Grab’s high-margin GFin segment.
Putting it all together, Grab is building an all-in-one food empire that is vertically and horizontally-integrated. No other platform in SEA even comes close btw…
GrabMart (GMV):
If GrabMart achieves 20% share of the total online grocery TAM and achieves 6-7% segment adjusted EBITDA margin, that alone adds up to ~$1.5B, which is >3x of Grab’s 2024 total EBITDA.
GrabMart (Ads):
As of now, Grab is only monetising 1.7% of GMV through Ads. $CART is at 2.7% and aiming for 4% long-term. Assuming ads achieves just a 3% take-rate, GrabMart Ads will be a $750M high-margin business by 2030. Assuming 60% EBITDA margin, we get a $450M EBITDA line added to the business.
That’s $2B in EBITDA by the end of the decade that is 4x of 2025’s guided EBITDA from GrabMart alone (!)
Steve Jobs on the most important job of a CEO
“The greatest people are self-managing. They don’t need to be managed. Once they know what to do, they’ll go figure out how to do it… What they need is a common vision, and that’s what leadership is. Leadership is having a vision, being able to articulate that so the people around you can understand it, and getting consensus on a common vision.”
Steve continues:
“We wanted people who were insanely great at what they did… and the neatest thing that happens when you get a core group ten great people is that it becomes self-policing as to who they let into that group. So I consider the most important job of someone like myself is recruiting.”
Today US #PPI data is released the day before CPI data.
PPI + CPI data have a very strong correlation.
PPI leading the way for CPI numbers historically.
Bullish PPI today = pre-rally CPI
Bullish CPI tomorrow = new ATHs
Recently, there have been many web3 AI projects focusing on GPU. Initially, I was excited, but after conducting research, I have some concerns. Feel free to comment and discuss:
1) Market Analysis 💰
The market capitalization of traditional Web2 AI companies, investing in GPU and infrastructure, has surpassed the total market capitalization of the entire cryptocurrency market. This suggests that the GPU power in web3 is significantly lower than in web2. However, it's not just about who has more computing power; AI training also involves software design. The complexity of AI algorithms in web2 may surpass that of web3.
2) Infrastructure Analysis 🏠
Traditional AI companies adopt centralized management, with massive equipment centralized in data centers, meticulously planned broadband networks, and backup power supplies. They utilize the latest equipment, such as Nvidia H100 servers.
Web3 projects operate in a decentralized manner, where platform providers reward equipment providers. Equipment may be scattered worldwide, with most providers only offering graphics cards like the Nvidia 4000 Series. Fewer providers can afford server-class equipment.
We know that gaming over a local area network (LAN) is faster than gaming over the internet (with distributed devices). Similarly, web3 GPU AI projects may lack speed advantages.
Efficiency-wise, web3 AI also falls short of web2. For instance, suppose the U.S. population is a centralized web2 AI company, and the EU population represents a web3 decentralized project. Even though the EU has a larger population (more devices), the U.S. has the advantage in overall economic strength and operational efficiency. Decentralization comes with latency and relies heavily on retail stability.
3) Economic Model Analysis 📊
High-quality Web2 AI companies can continuously raise funds, secure bank loans, and reinvest profits to expand operations, maintaining market influence and market share. Conversely, high-quality web3 AI projects inflate their valuations by leveraging others' equipment to increase token prices and sell for profit. Traditional industry equity structures mostly involve VCs, whereas web3 projects often have VCs owning most capital but holding only about 20% of tokens, with the team and project participants holding the rest. This leads to potential token dumping and instability if equipment providers can't bear expenses.
4) Conclusion 🧐
While web3 AI projects may not have an advantage in the GPU race compared to web2, I believe web3 can bring web2 AI data onto the blockchain, ensuring data security and decentralization for AI. It shouldn't solely rely on network computing power and a few devices to manipulate token prices.
Thanks for diving this far. Since your here I would appreciate if you can like and RT, also leave a comment and let’s discuss! 😆
#web2 #web3 #AI #GPU #LLM