MrRussell (@RodrigoPRussell) header
MrRussell (@RodrigoPRussell) profile picture

Partner @SumCap | ex-CRO & COO @xMoney_com (acquired by @MultiversX) | ex-VC @brpxcapital turned degen | Laissez-faire economics | 1/2 🦍🧬

Portugal
Joined February 2019
Photo by MrRussell (@RodrigoPRussell)Photo by MrRussell (@RodrigoPRussell)Photo by MrRussell (@RodrigoPRussell)Photo by MrRussell (@RodrigoPRussell)Photo by MrRussell (@RodrigoPRussell)Photo by MrRussell (@RodrigoPRussell)Photo by MrRussell (@RodrigoPRussell)Photo by MrRussell (@RodrigoPRussell)Photo by MrRussell (@RodrigoPRussell)Photo by MrRussell (@RodrigoPRussell)
MrRussell retweeted
Economist tier list:
815
930
219
8,484
843,381
MrRussell retweeted
the ai trade won’t bottom until market participants finally learn *not* to value companies based on fwd pe. a fwd pe of 40 is not necessarily expensive if earnings grow durably over decades. a fwd pe of 4 is not cheap if earnings last 18mo then go down 90%. a company’s value is the net present value of *all* future cash flow, not the net present value of *next year’s* cash flow.
56
26
9
336
38,664
Everyone argues about who has the best model. The real question is who can serve intelligence the cheapest at scale. Own the chip, own the cost curve. Google can undercut everyone on price and keep the margin Rivals rent compute and Google makes it. In AI, the lowest-cost producer wins This is the moat the market keeps missing, and it's still priced like an ad company. $GOOG
Google’s $GOOG performance-per-dollar improvements with each TPU generation is helping it offer among the lowest “cost per intelligence task”, with Gemini 3.1 Pro Preview costing an average of $1.30, or 51% cheaper than GPT 5.5 (high) and 76% cheaper than Claude Sonnet 5.5. Google’s TPU 8i, delivering up to an 80% performance-per-dollar improvement over Ironwood, could extend this further. io-fund.com/ai-stocks/google… $NVDA $AMD
3
66
$GOOG posted its first miss in ~4 years and the market is selling hard. I think almost everyone is missing the real story. Cloud did $30.7B, up 103% y/y, beating estimates by $2B. JPM says Cloud could equal Search revenue by late 2028. A second Google is growing inside the first one. The chips story is the one people keep missing. Google designs its own TPUs, and I wouldn't be surprised to see them lead in silicon too. Look at the whole stack: a frontier model that's finally winning, quantum leadership, its own compute, and data from Search, YouTube, Android and Chrome. Most rivals rent compute and buy data. Google makes both. Yes, capex is huge: 2026 at $195-205B, with 2027 "significantly" higher. That's an investment, not a cost. Compute will get monetized 100%. We're talking Google here frens. And the price: ~19x forward P/E at pre-market, against MSFT 22.4x, AMZN 24.4x and AAPL 36x. That's a discount for the most complete AI stack in big tech. The personal win here: added $GOOG onchain yday. Props to @backedfi, while the US market was closed. No broker, no waiting for the open. 24/7 access to equities is what crypto and onchain markets are for. Really short-sighted and uninformed market. 10T coded.
1
10
105,409
Replying to @chumbawamba22
setup: worst cross-asset undeperformance in crypto's history past few years driven by • brain drain to AI • lack of innovation post-defi cambrian explosion • gold exceptional once in decade deca-trillions rerating add on top 10/10 worst liquidation day for alts ever sentiment and positioning on crypto from insti & retail side at this bottom on altcoins are at multi year lows, maybe the lowest ever only one direction to go, up, in an asset class where it is heavily momentum driven ---- now on the flip side, for the first time in crypto's history, there are altcoins with • zero to little VC overhang • real pmf uniquely powered by crypto rails • regulatory environment finally leaning positive • everyone sitting on infinite unrealized gains from AI supercycle the industry is no longer spray and pray, and the coins that will receive bids are going to move extremely quickly
42
47
7
501
110,438
MrRussell retweeted
over 15 months of published code over 6 months on mainnet over 1.3B in assets We're seeing more and more LPs choosing Aave V4.
12
13
160
14,129
MrRussell retweeted
Aerodrome just announced they’re launching on Robinhood Chain. Just for reference, here is Uniswap’s DEX volume: all the red at the end of the chart is just RH chain. Btw, UNI is up 3x since the summer.
9
26
1
283
29,632
Teams aren’t running out of intelligence. They’re running out of tokens before the work is done. Soo we stopped guessing output length and started forecasting it. Available to all the builders now.
We’re open-sourcing our research on model output tokens. Usage budgets bottleneck teams shipping with LLMs. Too often, weekly limits arrive before the work is done. So we kept asking: can you estimate output length before generation starts? The result is Token Forecaster.
2
4
151
MrRussell retweeted
Everything you need to master in 2026 to get rich: • Using Astra 6 high for planning, low for execution • Grok Bot to manage day to day • Coding loops • Taking advantage of Codex's computer use • Herdr to manage tons of agents • /goal • Running local models • Karpathy's Autoresearch • Building an X audience • Creating videos • How machine learning works • How databases work (been using Convex) • Using API's • Training your own LoRAs • Elimination of doom scrolling • Most importantly: focus
128
117
13
1,826
128,636
Frfr.
there will be 10+ 0->1B+ crypto protocols formed over the next 12-18 months, mostly driven by RWAs moving on-chain + increased interest from talented developers realizing the immediate incentive alignment & made easier by constantly improving AI models
130
zero doubt in my mind this is going to be the best cycle we’ve had so far
595
480
99
6,827
268,891
MrRussell retweeted
Two protocols can run the exact same strategy and hand you completely different risk. The difference is tranching. A thread 🧵
1
2
10
837
MrRussell retweeted
Our team took home the win at Agents Day Braga. One day. One idea. A multi-agent debate simulation for prediction markets, finding cross-market hedges through emergent patterns of discussion. Proud of the team. More on this soon. Thanks @subvisual @LayerXlab @TalentProtocol for an incredible event.
Agents Day Braga is a wrap! 🕵️‍♂️ Thank you to everyone who showed up to work and to those who shared their knowledge through workshops: @0xdavidesilva on Agentic Development, Mário Alves and @thatCarlosM on Devs-GTM, and Simão Nogueira on cloning OpenClaw to run in the cloud for thousands of users. Congratulations to the winning project, Polycenas, on an amazing Reddit-style multi-agent simulation for Polymarket prediction markets. Thanks to @LayerXlab and @TalentProtocol for co-organizing. See you at the next one. #AgentsDayBraga #DevsPT
2
2
12
1,410
MrRussell retweeted
After 3 years and 3 hardforks, #Fusaka concludes the #Surge phase of the Scaling Roadmap with: - 📊 PeerDAS: 8x throughput, cheaper fees, more rollup space - 📈 60M gas limit: 100% increase YTD - 🟪 and 10 EIPs improving CL and fee markets Let's dive into what Fusaka entails 👇
3
4
1
15
1,938
Long live the doers, the wealthiest of all.
Wealth is the ability to make things happen.
1
3
246
MrRussell retweeted
Wealth is the ability to make things happen.
623
1,981
187
17,081
629,893
MrRussell retweeted
Plasma launched with 55+ protocols and $2.7B+ in pre-commit TVL, through partnerships with @binance, @ether_fi, @maplefinance, & @MidasRWA. A clear success, but what does the data reveal beneath the surface? Introducing the @Plasma Dashboard 👇 🔗plasma.sumcap.xyz
13
21
4
137
29,556