🇧🇷 15y Public Equities (ex-Head of Research @ top Brazil fund). Experience w/ Alternatives & Illiquids. Former Public Co. Board Member. Tech enthusiast.
Rio de Janeiro
- Tweets 3,094
- Following 2,036
- Followers 2,890
- Likes 8,573
Agentic commerce exterminating the 'Friction Rent'
Might be writing an article on this.
One thing I'm sure: the whole basket shouldn't move as a single factor. Now I'm digging what are the relevant questions to be answered or, at least, create a framework to work with them.
1
2
698
Igor Chalhub retweeted
The "system of record" SaaS thesis has become the "Merchant of Record" consumer thesis.
Time to update your priors.
1
4
347
Two generations of market participants never had to separate nominal growth from real returns, let alone think about the hurdle rate.
Now that risk-free Treasuries yield north of 5%, capital is no longer free. ZIRP is over, we're back to Macro 101.
Brazilians know better.
6
553
The emergence of proactive consumer agents is exposing a brutal, age-old truth: much of what consensus praised as durable pricing power and customer loyalty was misunderstood as a moat.
It was just a tax on human slack. We are lazy. And busy. Consumers have better things to do than spend 45 minutes on hold to save $20.
6
555
Muse altered the industry’s trajectory. Every major platform will now be forced to field a proprietary agent to defend against value expropriation and disintermediation. The defensive playbook will vary from walled gardens to strict Agent-to-Agent (A2A) handshakes designed to ring-fence proprietary data, margins, and customer relationships.
What is now unequivocal:
A. The harness is the product: Mass consumer adoption won't be gated by model availability, but by agentic scaffolding and execution harnesses (the OpenClaw / Muse paradigm). Once the harness solves task execution, consumer usage inflects vertically.
B. Frontier IQ is unnecessary for task completion: You do not need frontier-tier reasoning for day-to-day execution. Commodity "average-IQ" models perform exceptionally well once paired with a deterministic harness and the right toolchain.
C. A massive new vector for compute: The narrative that inference efficiencies would compress compute demand is wrong. Multiplying agent loops and autonomous workflows creates a step-function increase in token consumption.
Portfolio Positioning:
I still see zero durable moats forming at the application layer, only mass experimentation with CapEx at all time highs. With every usual suspect (Google, Amazon, OpenAI, Anthropic, etc) poised to ship competing agentic interfaces in short order, app-layer pricing power will get competed away.
Consequently, I am taking profits in $META. I owned it when consensus doubted the logic of building out proprietary compute and questioned incremental ROIC, while the near-term advertising monetization runway was obvious. Roughly +25% in 3 months and I'm good.
The market's mispricing has now shifted down the stack. The line every buyside is tracking is the capex 2nd derivative (accel/decel). Consensus is obsessing over "peak capex" and questioning cycle sustainability at the exact moment a blockbuster product architecture has emerged. And we just got a huge new demand vector (see "C").
Everyone is about to clone and deploy this agentic harness playbook and the aggregate compute demand required to run it is about to accelerate again.
Time to adjust the portfolio accordingly.
Random thoughts on the agentic shopping paradigm shift:
1. $AMZN: The initial reflex is predictably defensive, just throw up a walled garden around the store. Amazon will almost certainly run its own shopping agent, but the endgame is likely agent-to-agent negotiation: an external agent, e.g., OpenClaw (ops, Muse I mean) passes intent and constraints, and Amazon’s internal agent executes and settles the cart. Ultimately, physical logistics density and endless FBA SKU depth remain an unassailable moat. You can disintermediate the frontend, but you can't hallucinate the warehouse.
2. $SHOP: Seems like a structural winner. The long-tail merchant no human would ever browse, or even know existed, suddenly gets indexed and surfaced by autonomous agents scraping for exact specs. It levels the discovery playing field, unlocking massive incremental GMV for Shopify and its merchant base.
3. $GOOGL: The real structural loser here is search advertising. Agents don’t browse SERPs, care about SEO, or click Product Listing Ads (PLAs). Merely building a proprietary shopping agent would miss the point; Google needs to reinvent its core monetization rails from scratch, essentially building an auction architecture for machine-to-machine discovery (bidding on agent consideration sets). Even if Mountain View has the technical muscle to pull this off, the narrative damage and multiple compression will hit first. By the time the product ships, the market will have already de-rated the stock.
Thoughts?
7
8
134
33,739
Random thoughts on the agentic shopping paradigm shift:
1. $AMZN: The initial reflex is predictably defensive, just throw up a walled garden around the store. Amazon will almost certainly run its own shopping agent, but the endgame is likely agent-to-agent negotiation: an external agent, e.g., OpenClaw (ops, Muse I mean) passes intent and constraints, and Amazon’s internal agent executes and settles the cart. Ultimately, physical logistics density and endless FBA SKU depth remain an unassailable moat. You can disintermediate the frontend, but you can't hallucinate the warehouse.
2. $SHOP: Seems like a structural winner. The long-tail merchant no human would ever browse, or even know existed, suddenly gets indexed and surfaced by autonomous agents scraping for exact specs. It levels the discovery playing field, unlocking massive incremental GMV for Shopify and its merchant base.
3. $GOOGL: The real structural loser here is search advertising. Agents don’t browse SERPs, care about SEO, or click Product Listing Ads (PLAs). Merely building a proprietary shopping agent would miss the point; Google needs to reinvent its core monetization rails from scratch, essentially building an auction architecture for machine-to-machine discovery (bidding on agent consideration sets). Even if Mountain View has the technical muscle to pull this off, the narrative damage and multiple compression will hit first. By the time the product ships, the market will have already de-rated the stock.
Thoughts?
We are excited to announce we are partnering deeply with Muse to enable agentic checkout with Shop Pay on all Shopify stores, offering people an easy and delightful way to shop and check out with Muse.
29
14
207
75,611
NFLX downgraded to Underweight by Wells Fargo (PT cut $80→$57): 2H26 hours/sub tracking ~4% lower YoY, Top-100 Original viewing down >20%, 2027/28 EPS cut to $3.77/$4.52. Catalyst to watch: Jan-27 viewership report.
3
1,303
Igor Chalhub retweeted
That was a deliberate and conservative decision made by Dario to pace datacenter capacity according to revenue visibility in order to protect Anthropic’s balance sheet.
Once Opus 4.5 and Code were blockbusters, that compute SxD became deeply unbalanced.
Now Anthropic adjusted course, but it takes DC construction to finish (18-36m).
On the other hand, OAI went all-in and put BS at risk. But it worked.
1
2
507
"Phil Fischer noticed that close to 80% of his best ideas actually came from other investors... the best filter of information is other smart people."
Who do you follow at X that is high alpha signal?
2
6
913
"You were hired for your opinions, not your knowledge." - Michael Moritz
Always great to learn from masters of the craft. It does not matter if it is VC, PE, or listed equities.
Consume information broadly.
Process it deeply.
Communicate it narrowly.
Act upon it.
Sir Michael Moritz is a true legend. My first week at @sequoia, my first partner meeting, and I feel completely clueless. The discussion revolves around a chip design company, which I literally know nothing about. Then and now. Michael asks me what my point of view is, and I reply that I know nothing about it. At the end of the meeting, Michael approached me and said softly, "You were hired for your opinions, not your knowledge," then walked away.
My first VC learning. One of many I learned from Michael.
1
6
1,119
Uma das coisas que aprendi é que mais relevante do que ser expert no jogo, é escolher, com muita atenção, o jogo a ser jogado.
3
497
Alguém precisa falar o óbvio.
"A bolsa brasileira está barata" virou a tese mais preguiçosa do mercado.
Nos anos 1990 e 2000, bater perna em fábrica e entrevistar ex-funcionário gerava alpha. A assimetria de informação era brutal e a concorrência, escassa. Hoje, isso é a linha de base. Todo mundo sabe que o múltiplo está deprimido; e a precificação do óbvio não gera retorno excedente.
Pior: a tese ignora a clássica armadilha do valuation ("value trap"). Se o múltiplo está baixo, há uma razão para isso. Sem um motor de crescimento sustentável e uma liderança competente, o investidor não está fazendo um investimento perene; está apenas operando um trade tático do ponto A ao ponto B.
O ponto central que a narrativa do "múltiplo atrativo" ignora é simples: o Brasil historicamente não entrega Equity Risk Premium (ERP). Correr o risco da renda variável local raramente compensou o carry do juro real. Quando a bolsa brasileira anda, o movimento raramente vem de tração econômica orgânica ou crescimento secular de lucros, mas quase exclusivamente da compressão cÃclica da taxa de desconto. Na prática, compra-se juro alavancado, e não participação em empresas de crescimento sustentável.
As campeãs da B3 já fizeram o dever de casa: executaram seus planos estratégicos, otimizaram margens e atingiram seu market share de equilÃbrio. O problema é que não existe heroÃsmo microeconômico que compense uma macroeconomia estagnada. A performance individual dessas companhias colidiu com o teto do PIB.
Alocação de capital é, antes de tudo, sobre a escolha da qualidade do aquário onde se pesca. Infelizmente, o aquário brasileiro está sem manutenção há décadas. Até tem peixe, mas estão sufocados pela falta de oxigênio institucional e de produtividade.
O aquário global, em especial o americano, atrai muito mais pescadores e uma competição acirrada, mas a água é limpa, o dinamismo tecnológico é real e o ecossistema renova sua biomassa diariamente.
Fazer um stock picking brilhante em um aquário de águas turvas e ar rarefeito continua sendo uma estratégia de retornos medÃocres e cÃclicos.
Nadar a favor da maré sempre será uma escolha mais inteligente.
$EWZ #IBOV
6
3
57
8,700
Key quotes from the recent $NU podcast on Global Expansion
Full episode: piped.video/MysGC2-Fj-Y?si=MeS3…
2
3
578
The data delivers a sobering reality check for the 'buy-and-hold' crowd.
Between fund management and personal investing, I’ve looked at thousands of businesses.
One uncomfortable reality is that 99% of them are just companies - good people with a reasonable product, doing the best they can. This is absolutely admirable and important to the economy.
Maybe 1% are truly special. And usually, that’s because they have the courage to focus and also do the fundamental things exceptionally well.
Unfortunately, this also makes it hard to explain exactly what makes them special. There is an element of just knowing it when you see it.
Spending your time learning to identify these companies is the highest-return activities an investor can pursue.
1
1,065
US diesel went through $6 a gallon this week and the Senate Majority Leader told reporters he is open to exploring an export ban, so what a diesel shock costs a named operator stops being hypothetical.
$AMZN worldwide shipping costs grew 19.2% in Q2 against 17% paid-unit growth, ending 15 consecutive quarters in which units outgrew shipping cost.
The reading everywhere is that fuel broke Amazon's cost engine. Olsavsky's (CFO) own sentence on the call says the engine ran again this quarter and that fuel ran faster.
Amazon spent the last six years rebuilding its fulfillment network around regional self-sufficiency, and the premise of that rebuild is fewer miles per package and increased client NPS. A record diesel crack is the first environment that tests it.
1
1
521
Three ways this is wrong.
The $4.85 fulfillment fee is a price, and a price can sit below the cost of serving it, so if Amazon prices FBA below cost to hold third-party volume, the gap against carrier freight is measuring a subsidy and the density read fails. I have no cost-side evidence in either direction, which makes this the weakest link in the argument.
Fuel's share of the shipping-cost line is undisclosed and I used 8%. Above roughly 14% the same growth gap becomes a materially larger dollar figure and the 36 bps framing weakens with it.
Advertising at a ~$79B run rate is what makes the fuel line absorbable, and on August 31 the FTC and 22 state attorneys general sued that exact business, alleging Amazon inflated prices in the Sponsored Products, Sponsored Brands and Sponsored Display auctions and seeking an injunction against the pricing mechanism itself. The separate FTC marketplace case goes to trial in October, the same month Amazon prints Q3. The two risks therefore compound conditionally rather than adding, and elevated diesel arriving into an impaired ad engine is the pre-mortem that damages the long-duration case.
Against myself on the guide: it reads as up 29% to 52% against $17.4B a year ago, but Q3 2025 carried $4.3B of charges from an FTC settlement and severance, so the clean base is $21.7B and the midpoint implies 13% growth on 11% revenue, which is a flat margin. The sequential step-down from clean Q2 is real, and the Prime Day shift into Q2 accounts for roughly half of it while fuel accounts for about 15%.
1
1
115
Q3 closes on September 30 and the crack spread only set its record on September 3, so the quarter carries about one month of the spike against three months of volume. That makes Q3 an early read. Q4 is the test, because a crack that holds lands a full quarter at peak-season volume, and the household fuel bill reaches its own year-over-year peak in the same weeks.
What would move me, with current levels stated.
→ Q4 shipping-cost growth exceeding paid-unit growth by more than 5 pts, against 2.2 derived in Q2 and roughly 3 implied for Q3 on one month of the spike
→ FBA surcharge raised above 6.0% before year-end, since that is management revealing a drag several times what I have modelled
→ Q3 operating income below the $22.5B guidance floor, which is the early read on a quarter carrying one month
→ NA retail operating margin below 6.0% in Q4
→ Paid unit growth below 10% in Q4, against 17% in Q2 and 12% in the same quarter a year earlier. Units are the demand read here because Amazon discloses no GMV, so a falling average ticket on its own measures mix
One caution on how that fifth trigger will be reported. A regressive fuel bill rotates the basket toward lower-ticket items and toward third-party sellers, and third-party sales are booked net, so a GMV dollar moving from 1P to 3P cuts reported revenue by roughly half while lifting gross margin. A Q4 revenue miss that arrives with units intact and gross margin up is that rotation, and it will read as a demand break.
The input I would most like corrected is fuel's share of the shipping-cost line. If you have a fleet, miles and MPG build of Amazon's Class 8 line-haul exposure, that number decides most of this thread.
84