CPX Continues Independently, Partnering with Groq to Accelerate AI
Getting some clarification, but CPX is not dead will go forward on its own product path along side Groq. Both have separate, but very specific jobs to accelerate AI.
2027 Market Nears $600B, Networking Over $60B
2027 more likely to be closer to $600B. Networking alone could be north of $60B.
Optical Interconnects Will Dominate Over Copper for Scaling
It does seem copper will be an option in Feynman, BUT, I'd bet the mix shifts heavily in favor of the optical options in this timeline. A comment was made in a session I was in yesterday that once you...
Agentic CPUs Demand Top Single‑Core Performance, New Architecture Needed
As always, solid interview by @benthompson with Jensen. Lots of nuggets, but one key for me is agentic CPUs requiring the best single core performance. Seems likely agentic CPUs will require a different architectural approach. https://t.co/uovoNuv1fg
NVIDIA’s Visual Guide to Scaling Up, Out, Across
In case it helps here is a visual on how NVIDIA describes scale up, out, and across. https://t.co/BVhQbE17Qk
Integrating Groq Keeps In‑House, Not Outsourced
Jensen is articulating what my main concern with Groq was in that as an independent solution why would I outsource all my inference. By integrating it as a part of the solution, it's just disaggregated but not 100% outsourced,...
NVIDIA's Co-Designed System Delivers 35× Efficiency Boost
The new system of compute from NVIDIA. Extreme co-design = 35X more throughput per MW. https://t.co/Z6JtDKckOh
NVIDIA's Massive Installed Base Fuels Ecosystem Strength
Jensen is emphasizing something he has said a few years now on the strength of the NVIDIA ecosystem. Massive installed base and architecture compatibility.
Nvidia Can Tap Every AI Funding Stream
Bear case aside, $NVDA can pull from almost all of these dollar buckets for AI infrastructure. https://t.co/5HKeoRjnJM
NVIDIA's Expanding Stack Inflates Total Infrastructure TCO
One of the more interesting things I'm curious about is how, as NVIDIA expands its infrastructure offerings, all new infrastructure competes for GW. The story has to be about the compounding TCO of the full solution, beyond GPU racks.
OpenAI’s Unified Platform Powers Three Distinct Engines
On the back of the report on Anthropic, publishing our full growth thesis on OpenAI. @DiligenceStack OpenAI: Three Engines, One Platform https://t.co/9inLX4EVvr
Future AI Agents Will Require Vastly More CPU Power
I would also add that agents will need a lot more CPUs.. Via MS note on software in an agentic world. https://t.co/eJpBqngKLu
Apple May Shift to Device-as-a-Service for Education
This got me thinking, this could be how Apple starts to offer devices as a service to things like education, perhaps enterprise someday.
Nvidia Controls 86% of AI Infrastructure Market
In the AI infra semiconductor TAM ~$1.4T estimated for 2030 there are 18 major sub TAM buckets of varying sizes. $NVDA plays in 12 of those 18 and that 12 $NVDA plays in = ~$1.2 T which is 86%...
Vertical Rack Designs Make Optical Interconnects Essential
This is when optical will move to scale. Feynman concept rack from CES, blades stacked vertically, not horizontally. Once rack designs move to this (may be training racks only), then optical becomes a must. https://t.co/CR2fBiX5kH
Few Know I Own Multiple Apple Devices
No. iPad, I think one version of watch if not more. Anyway, not the first time, just very few knew :)
Supply Chain Power Concentrated; Small Players Will Struggle
Ahead of GTC, since $NVDA is a name on this list, publishing this report on the power a few wield in the supply chain that those without that scale will struggle in securing supply for their demand. Masters of the...
CoreWeave Sued for Overstating AI Capacity and Hiding Provider Risk
Very interesting development as a lawsuit was filed on $CRWV. Using a lot of the angles the AI bears have been concerned about with "optomistic commentary." Will be interesting to watch. The core complaints are: 1. CoreWeave allegedly overstated its...
Mac Mini Hosting for Agentic AI Benefits MacStadium
Didn't have Mac Mini's as being the host computer for agentic AI on my bingo card. Looks like they are hosting a Mac Mini for you as part of the service. Maybe a good play for @MacStadium ?
Apple’s GCD Update Boosts Efficiency on M5 Max
I think Apple has updated Grand Central Dispatch to squeeze out some of these efficiency gains on M5 Max. Sitting here staring at core utilizations by workload and there are some differences from prior Max/Pro silicon.
Copper Stays; Hyperscalers Diverge on Optical Mix
I have zero doubt copper hangs around for a long time. As @theaustinlyons says "copper if you can, optical if you must." The hyperscaler divergence is the key point to watch here. Each will go down their own path, mix...
Timing Tracker Reveals Why Optical Advances Awaited
Optical has always been about timing. Lots of folks were right "banging the table" but anyone who has tracked this space knew it would happen, and why, but when was the question. That's why in this report, I outlined...
Apple Poised for Massive Influx of New Ecosystem Users
Exactly what I dig into in this report. Apple is very likely to see one of the bigger cycles of net new customers into the Apple ecosystem. https://t.co/JOdLgVdVhl
MacBook Neo Overkill for Most Notebook Users
I'd wager Macbook Neo has more than enough compute power for 70-80% of all notebook users.
First in Maui with MacBook Neo, Amazing Speakers
MacBook Neo in Maui. I may be the first person in all of Maui rocking MacBook Neo lol. Also the speakers on this. Incredible at this price https://t.co/iXGKFqWCUe
Apple Emerges as Full‑Stack Compounder with Clear Growth Strategy
I have yet covered Apple in the Dilligence Stack, so felt this week was as good as any. Apple has a lot of advantages set up for themself and now the execution strategy for growth is becoming more clear....
Neo Matches MacBook Air, Beyond Just Light Tasks
The other distinction with Neo that will be interesting is it is capable of more than "light work." There is nothing I can't do on it I can't with my Macbook Air. The only audience this won't cater to...
A18 Pro on Neo Peaks at 8W CPU, 14W GPU
Max wattage I could push A18 Pro to on Neo was CPU ~8W and GPU ~14W I may be the only one, but was very excited to see A series processor wattage under workloads for the first time. https://t.co/0cQbcNZFT0
Notebook Delivers True Workstation Performance with Efficient Power Management
It’s a true workstation in a notebook. The efficiency story and the way power is managed for not just the S and P cores but also the CPU/NPU tile and the GPU tile is impressive.
Apple’s Mac Growth Spike Fueled by Neo Expansion
I have LONG argued, Mac is Apple's greatest short term growth lever. By adding Neo, they just added potential TAM grab of 50m units a year from that price band $600-800 Apple sells ~25-30m Macs depending on the year. ...
GeekBench Leaks Confirm Apple’s Exceptionally Capable SoCs
All these GeekBench leaks just double emphasize the point that Apple designs remarkably capable SoCs.
Nvidia’s Compute‑Revenue Model Proven by Platform Economics
I'm glad Jensen is doubling down on the compute=revenues point. I dove into the full economics of why in this full report on how $NVDA platforms enable those economics. https://t.co/wrL7SktAYA
NVIDIA's Pixel-Friendly Architecture Also Excels at Token Processing
I wrote a report on this a while ago that the key thing that moved in NVIDIA's favor is the same architecture that is perfect for processing pixels is the same that is perfect to process tokens. This is from...
Apple Aims MacBook Neo at Sub-$800 PC Market
In case folks need the context, the majority of consumer PCs sold annually (~120-140M depending on the year) are in the sub $700-800 price tier. 95% or more Windows-based. In markets outside US price tier is under $600....
Apple’s Superior Swap Makes 8 GB RAM Sufficient
A main reason why the 8gb of RAM won't be an issue for the type of workloads/web browsing, etc., folks in this price band do regularly.
Apple Eyes Intel Foundry for Disaggregated Chip Design
One step closer to a fully disaggregated design. Which is inevitable at some point. Also, where we think Apple has an opportunity with Intel Foundry via EMIB/T.
Agentic CPUs Could Rescue Intel Amid Foundry Push
The return of the CPU will, interestingly, benefit Intel in a few ways. If they pull off my foundry thesis here, the rise of agentic CPUs could be the savior. https://t.co/dKqm16myKH
Tech Brands Overhype ‘Super’ Specs in Marketing
I know it's all marketing, but NVIDIA calls their GPUs super chips, really super all the things, and now Apple with super cores. Super memory, super storage incoming? 🤣
Apple Adopts True Chiplet Architecture for Independent CPU‑GPU Scaling
Apple is doing a much closer true chiplet design with these than with Ultra which is just two bonded SoCs. This separates the CPU/NPU from the GPU which brings some interesting options to scale both independently of each other. Likely...
Nvidia Co‑optimizes Every Vendor Component, Not Off‑the‑shelf
Worth pointing out, no third party vendor $NVDA works with will be an off the shelf implementation into their designs. Everything will be co-optimized and intentionally designed into their platforms/systems. Relevant when thinking about the $LITE and $COHR plays.
Memory Crunch Threatens Custom ASIC Roadmaps and TCO Edge
With the Meta custom silicon, and recent Google TPU JV (selling TPU hardware), the dynamic for custom ASICs has completely changed with the memory crunch. Need to reconcile how not getting access to leading edge HBM will impact their...
NVDA Secures Memory Supply Through 2027
$NVDA buying up all the memory. This makes more sense now that we realize they have secured most, if not all, they need for 2026 and some of 2027. https://t.co/Ll2W3Fcc35
Nvidia’s HBM Grab Mirrors Apple’s iPod Storage Monopoly
Been hearing via supply chain friendlies, NVIDIA has secured most of the HBM, making it scarce for competitors. Reminds me of iPod days when Apple secured all the storage as others were trying to compete with them in MP3 players.
AI Accelerator Market Needs Diverse Solutions; Intel Takes Step
The AI accelerator/compute market is not a one-size-fits-all market. There will be a host of solutions and a variety of workloads. While Intel has yet to show an AI accelerator solution, this is a step in the right...
Memory: From Overlooked to Industry’s Hot Topic
Went deeeeepp on the memory industry in this report. No one used to want to cover memory, and now it is all the industry/supply chain wants to talk about. Some juicy stats 😀 https://t.co/QSHAd1OxK2 https://t.co/dbC4xNxVte
COT Shift Signals Changing Custom ASIC Customer Landscape
Important commentary here on COT on the $CDNS call last night. Called out this trend last year. It will be important for $AVGO $MRVL $MTK customer dynamics in custom ASICs. https://t.co/lgWuwvj4oB
Hybrid AI Wins; System-Level UI Integration Leads
Good points from @benthompson @stratechery today. Hybrid AI is the way, and those OS/UIs that manage that best and integrate agentic at the system level are best positioned. https://t.co/b2PY6FuSgR
Confident Helios Shipments Will Arrive on Schedule, Confirmed
There was zero doubt in my mind that there were any delays to Helios, and the first shipments would be on time. Glad @AnushElangovan confirmed it.
Market Can Only Meet Half to Two‑Thirds of Demand
Morgan Stanley hosted $MU CFO. "the gap is significant" "commentary that the state of the market is such that it can only support 50% to 2/3rds of key customers’ demand," https://t.co/NLNcHwM3ri
Semiconductors Shift From Volume to ASP‑Driven Value
Modeled a range of scenarios, but showing mid-bull to make a key point. For the first time in semiconductor industry history, value creation is primarily an ASP story not a volume story. https://t.co/IoHFb65Gmq