So following these principles here is the architecture that falls out and it’s quite different than what is thought to be optimal today.
Starting with the memory: LPDDR6 tops out around 14.4 Gb/s per pin. We would want to as many channels as the die’s edge can carry. If we can fit 24 channels, that’s 576 data pins, and you get ~1 TB/s per chip with 192 GB of storage. Same as the b200 in size but about an a100 in bandwidth.
But remember that LPDDR access latency is ~100 ns which is about the same as HBM. So streaming weights from LPDDR costs you nothing on latency and we just give up per package bandwidth, and remember our pipelining already collapsed that requirement.
Now work backwards to compute. Tokens/sec = ~aggregate BW / bytes touched per token, and at batch 64 in FP4 every delivered byte wants ~256 FLOPs. 1 TB/s × 256 = ~260 TF of FP4. Size ~280 TF for some headroom and stop! That’s all we need, the rest of the alpha is in latency tuning not flops.
Now process node, again using first principles thinking. On TSMC N6 that MAC array would be about 30 mm^2. The rest of the die goes to SRAM buffers, 24 LPDDR PHYs, and 56 lanes of plain 32 GT/s NRZ SerDes PCIe 5.0 transceivers, nothing exotic. Because we didn’t provision more flops we can provision what we really need which is ports. If we do 7 ports × x8 we can get a full mesh across 8 chips with ~224 GB/s any to any ops, sized for 8 KB activations instead of 100 MB gradients. Approximate Total Area: ~144 mm^2, ~180W with the DRAM. Fucking Mobile SoC level economics.
And we get away without a crazy process node. N3 shrinks your logic, but that’s actually the one thing this die barely has. SRAM stopped scaling and the PHYs and SerDes are analog so they don’t shrink at all. An N3 port shrinks ~20% of the floorplan, and saves maybe 10-15W, but doubles the cost on a small die that already has beautiful *chefs kiss* 👨🍳💋 yield.
Now do the node math:
8 chips (b/c full mesh w/ 7 ports) = 8 TB/s aggregate BW, 1.5 TB total of memory, and ~1.44 kW. A B200 inside an NVL72 is 8 TB/s, 192 GB, ~1.67 kW all in. So we are basically neck and neck on BW with a b200 with slightly lower power because we have more chips connected with quicker interconnects allowing them to act as a single unit.
If we scale to the rack level at 72 nodes, that 576 chips, 576 TB/s which is again the same aggregate bandwidth as a GB200 NVL72, so approximately the same tokens/sec by construction but with ~104 kW of power against ~120-130 and 110 TB of memory against 13.8 TB of the NVL72. So in summary 4.5x less compute silicon, zero HBM, zero CoWoS, zero leading-edge nodes and still token/watt ~20% better.
So what’s stopping everyone from doing this? The number I told you hold whether you want to believe it or not and get even more competitive with longer reasoning chains or agentic workloads because as you scale those the all reduces now sits on the critical path of every token and on a 30 ns link with a full mesh that’s orders of magnitudes of token throughput improvements over Nvidia’s stack. Counterintuitively the link latency, not the chip architecture, decides your memory, logic, and everything else!
Thanks for reading if you made it this far.
How to find leaders before they go parabolic:
My favorite swing trades are in stocks I call the "grinding leaders that can become parabolic leaders".
Grinding leaders = stocks grinding up <10% a month across longer timeframes (months)
Parabolic leaders = stocks making a 50%+ parabolic move within a very short timeframe (1-2 weeks)
Join strength during the grinding phase, pre-parabolic move...here's how to find them:
Use a screener (I use finviz), and filter by:
- price above SMA20
- price above SMA50
- price above SMA200
- performance month: up 5% to 15%
- performance quarter: up 20%+
- RSI(14): 55-70ish (optional, but avoid already insanely overheated names above 80 RSI)
- average volume: over 500k
- market cap greater than $300M (or $2B+)
Optional catalyst layer...this is the real alpha!!
After the scan, I manually look for:
- hot theme exposure (ai, robotics, defense)
- earnings acceleration
- recent news
- sector rotation
- big partnerships
Few names sticking out to me:
$SATS
$IRDM
$FLY
$EBAY
$IBKR
...stay tuned for deep dives.
💙Luc
I promise this will be the best 20 min you spend today! Robotics: Endgame, the sequel to my last year's Sequoia AI Ascent talk, "Physical Turing Test". I laid out the roadmap for solving Physical AGI as a simple parallel to the LLM success story. Be a good scientist, copy homework ;)
And stay till the end, more easter eggs and predictions for your polymarket!
00:30 DGX-1 origin story at OpenAI, I was there in 2016 signing with Jensen and Elon. Heading to the Computer History Museum!
01:42 The Great Parallel
03:31 Robotics, the Endgame
03:39 Why VLAs fall short
04:32 Video world models as the 2nd pretraining paradigm
06:09 World Action Models (WAM)
07:46 Strategies for robot data collection and the FSD equivalent to physical data flywheel for robot manipulation
11:06 EgoScale and the Dexterity Scaling Law we discovered recently
14:00 Physical RL: bridging the last mile
15:39 DreamDojo: an end-to-end neural physics engine for scaling RL in silico
17:00 Civilizational Technology Tree and my predictions for the near future. Spoiler: it's closer than you think.
Thanks to my friends at Sequoia for inviting me back to AI Ascent this year! I had a blast! Last year's talk is attached in the thread if you missed it.
The 9/21EMA crossback continues to be one of my favorite areas for momentum returning to a leader.
You just needed to recognize the character change and catch a piece of the trend.
+50% on the underlying in 12 trading sessions, right off the 9/21EMA crossback.
Chart: $MU.
⛽️ The West hasn't felt the full force of the oil shock yet, but the ripple effects are already visible
📉 This chart reveals a critical insight: The oil shock isn't initially felt in prices. It's first felt in the barrels that never reach their destination.
🚢 The Blue Line: Global Oil on Water
This represents oil currently being transported by tankers between producers and consumers.
When this line suddenly plunges, it indicates a serious disruption to the physical oil supply chain. Simply put, the oil that should be on the water is gone. This is always where the trouble begins for the physical oil market.
🛢 The Red Line: Commercial Crude Oil Inventories in Europe and the Americas
Interestingly, this line hasn’t dropped significantly yet. This shows that Western nations are currently relying on existing reserves, alternative routes, and emergency management to delay the main impact.
Think of it like a dam where the inlet water has been cut off, but the city doesn't go thirsty immediately because the reservoir is still full.
🧠 The Hidden Danger
Many believe a full-blown crisis only occurs when Western gas stations go empty. But the oil market doesn’t work that way.
The barrels that aren't on the water today won't be in the storage tanks tomorrow.
This delay is what makes the crisis so dangerous. The market may appear normal for a time, but underlying pressure is building.
⚡️ The Domino Effect
If the red line (inventories) also begins to fall dramatically, the situation will escalate from a geopolitical concern to a real, tangible shortage.
This won’t just impact oil prices. It will cascade through:
⛽️ Gasoline and Diesel Prices
🚛 Transportation Costs
🌾 Food Inflation
🏭 Production Costs
🏦 Central Bank Decisions on Interest Rates
The shock is on its way. The data shows it. It's only a matter of time...