Revolver Ocelot retweeted
I suggest everyone interested in Nvidia to read these slides posted on their site...it is interesting
Link in replies š
7
13
78
9,973
Revolver Ocelot retweeted
Mike. With all do respect people donāt subscribe to a Substack or research service to learn how to ātax loss harvestā
Just admit it and take the L and move on
@kakashiii111 has done excellent work on Nvda throughout the years, this is just another example of where he caught something early.
I suggest subscribing to my substack where I recently discussed closing my Nvidia shorts for a strategy called tax loss harvesting.
4
2
24
3,905
Revolver Ocelot retweeted
Itās here and on Instagram in a highlight and sway.co/ElizabethBarcohana
1
2
170
Revolver Ocelot retweeted
There is literally a map of human shit locations in your state and this is what you're doing? Amazing
111
419
18,563
248,618
Jensen was done selling GPUs along time ago.
First came the GPU.. then CUDA.
then network
then generative ai
then sovereign
then agentic
then physical
more software, open source, financial lending mechanisms for LPS
Catch me if you can $NVDA
2
8
59
5,853
Revolver Ocelot retweeted
The Intel 14A discussion is quite funny. You have two distinct camps
1) People who thought 14A was a N2P competitor (Asian based investment banks still here). So the news of within 5% PPA of A14 is shocking and dismissed. (They look at just HD/HP density ignoring everything else)
2) People who thought it would be firmly ahead of TSMC. Aka drinking a bit of the Pat koolaid.
Meanwhile you can just be happy that it's highly competitive with TSMC A14, not having the teething problems of 18A, and will be produced in the US 2 years before TSMC brings A14.
If you value the product design at a 20x PE ratio (aka assuming high single digit rev growth, which is too low IMO but just for comparisons here).
You are buying a Foundry at 2.5x P/B who has a process comparable to TSMC in PPA and having an excellent yield trajectory.
This is 80% cheaper than TSMC, which has a Taiwan discount on it's assets already.
$INTC $TSM
8
14
139
13,983
Revolver Ocelot retweeted
Evercore ISI: Power + thermal content per rack by GPU generation
$NVDA
7
17
80
11,485
Nvidia $NVDA reinstated as Top Pick in semis at Morgan Stanley
Morgan Stanley is reinstating Nvidia as the analyst's Top Pick in semis after having recently hosted Nvidia CEO Jensen Huang and CFO Colette Kress for investor meetings.
The biggest takeaways include the diversity of the customer base, growth within traditional hyperscale and frontier labs and that the strategy is to drive up revenue per GW substantially, to ensure that Nvidia solutions continue to drive the most tokens per gigawatt.
Recent developments in land/power/shell, financing, component constraints, and agentic "all play to their strengths," says the analyst, who has an Overweight rating and $300 price target on shares.
3
13
124
8,864
Up to 8x faster, and now you know why ā”ļø@openai GPT-6 Astra Ultrafast runs on NVIDIA
Read the story: nvda.ws/3W2bZ5H
131
106
1,284
461,749
Revolver Ocelot retweeted
Nike ALWAYS has some tremendous positive moments that lure you in but the aggregate outlook is terrible and, arguably, worsening...
399
69
1,478
537,534
Revolver Ocelot retweeted
Iām so confident in my $NVDA thesis that Iāve closed my shorts. Read more about how my subscribers are utilizing tax loss harvesting in my substack.
1
1
5
114
Revolver Ocelot retweeted
Agentic AI is changing the demands of AI infrastructure.
At #NVIDIAGTC Berlin, the leaders and engineers who design and run #NVIDIADGX are sharing their playbook. Explore how the DGX platform improves AI factory efficiency, speeds deployment, and prepares infrastructure for future workloads.
Add these sessions to your schedule now: nvda.ws/4jyXtMD
7
18
126
4,476
Revolver Ocelot retweeted
Sourcing compute has historically been slow
Last night we had a large block of Blackwell GPUs hit our marketplace. Within hours, multiple qualified off-takers had placed bids
We're working to make our platform the fastest way for you to acquire the compute your business needs
3
8
1,011
Revolver Ocelot retweeted
GF Securitiesā Jeff Pu:
$INTC Intel takeaway: this is an EMIB-T packaging win.
Reported facts on MediaTekās Google TPU v9 work (Humufish / Triggerfish):
- 100% EMIB-T packaging path
- Small silicon bridges and TSVs replace a large CoWoS interposer
- Substrate yields are around ~50% and improving
- Small-volume substrates mid-2027, mass volume 4Q27
- multi-million units
Essentially the logic die can stay at TSMC and Intel still gets paid. EMIB-T is the attach point - external advanced-packaging revenue, a second source while CoWoS stays tight, and a scale test of IFS with a hyperscaler program.
$GOOGL
Summary of MTK (2454) Update - 100% EMIB, 100% Share in v9
ā¢Street-high TPU estimates ā raised TP to NT$6,650 (from NT$6,420)
ā¢MTK fell 7% yesterday on noise around potential v8t delay, EMIB/substrate supply issues & v9 share-loss fears ā we take the opposite view
ā¢v8t/Zebrafish remains intact; demand well above 2027 CoWoS allocation of 180k, suggesting extended life cycle vol.
ā¢EMIB-T / substrate yields & supply tracking better than expected, at ~50% currently. Yield rate in every stage in report
ā¢EMIB Substrate delivery in small-volume ~100k mid-2027 ā mass-volume by 4Q27
šÆ 100% EMIB-T for v9 (Humufish, Triggerfish). on-track substrate yields + suppliers dedicated support. Also no reason for TSM to be a back-up or 2nd source.
šÆ 100% market share in v9 (Humufish for training, Triggerfish for inferencing). We expect no other v9/2nm projects during the time period
ā¢Life-cycle volume for v9 expected >5m (vs bears at >2m)
ā¢2nd ASIC project (likely SpaceX ASIC, not AI6) more visibility expected in 4Q26
ā¢Datacenter ASIC forecast raised to NT$2.8bn / $18.6bn / NT$75.7bn
ā¢Near-term watch: engineering tape-out in Nov (passing bar not high; chipset testing at TSMC largely on track)
ā¢For v10: still expect MTK to win major SKU Icefish (AMD+MRVL or AVGO competing for Merope)
#TPU #MTK #AVGO #AMD #MRVL #v9
#intel #intc
4
6
49
5,717
Revolver Ocelot retweeted
Cheaper AI can still mean a bigger compute bill.
Persistent agents can add planning, tool calls and retries. Our DevDay note asks what that means for NVIDIAāand what evidence the bull case still needs.
I'm long $NVDA. Free to read:
newsletter.bepresearch.com/pā¦
2
9
3,131
