Jeremie Eliahou Ontiveros รีทวีตแล้ว
First-try Gold, team smaller than a single hyperscaler standup - join us.
Fluidstack@fluidstack
Small team, big W. Join us to turn bronze tier infra into a gold tier cloud: careers.fluidstack.io/jobs
English
Jeremie Eliahou Ontiveros
163 posts
Jeremie Eliahou Ontiveros
@JeremieEO
Tracking hyperscalers, datacenters and energy infrastructure at SemiAnalysis. opinions my own https://t.co/le84JoyVsI
เข้าร่วม Mart 2022
523 กำลังติดตาม1.4K ผู้ติดตาม
@institLPGP @SemiAnalysis_ We wrote a deep dive on 800VDC about 10 months ago for institutional clients. Flagged some key beneficiaries which are all up nicely since
The way Bloom benefits from 800VDC is much more nuanced!
English
just gunna leave this here
from ubs analyst on kyber 800 volt architecture
@SemiAnalysis_ , skate to where the puck is going
$be
Ben Pouladian@benitoz
Bloom Energy just dropped a bombshell on their Q4 call. Their fuel cells now power absorption chillers using waste heat cutting data center electricity usage by 20%+ All is 800vdc No grid. No HFCs. Free cooling. Power + cooling from one platform. $BE 🔊 Listen to this 👇
English
Jeremie Eliahou Ontiveros รีทวีตแล้ว
Semi-Analysis's @JeremieEO on @tbpn last week:
"2025 was an insane year for the (data center) leasing market, and everything points to 2026, 2027 just accelerating."
Towards the end of 2025, Applied Digital $APLD, Cipher $CIFR, and Hut 8 $HUT signed the most attractive large scale (e.g. 200MW+) leases with investment grade hyperscaler counterparties to-date..
The incremental 830MW of power that Galaxy Digital $GLXY received final ERCOT approval on yesterday seems to have arrived at a fortuitous time..
English
Jeremie Eliahou Ontiveros รีทวีตแล้ว
Happy Friday. On today’s show:
- @bhorowitz (a16z)
- @arampell (a16z)
- @DavidGeorge83 (a16z)
- @jkhamehl (a16z)
- @JeremieEO (SemiAnalysis)
See you on the stream.
TBPN@tbpn
a16z's Size, Ben Horowitz Joins, AI needs a (Steve) Jobs, Tim Cook's Retirement, WSJ Mansion Section x.com/i/broadcasts/1…
English
@CapArtichoke @dylan522p @fundaai Our first mention of GCP's TPU upside was August 7th. In early September we quantified revenue growth in 2026/27 and called out strong acceleration.
Our first AWS note quantifying growth / calling for Accel was Aug 5th. AMZN has well outperformed Meta, MSFT, ORCL since then.
English
@dylan522p @fundaai Do you think you helped anyone buy AMZN/GOOGL/MSFT?
You do realize the entire world owns them yeah?
Google’s stock went up 50-60% in 2025 way before anyone mentioned the TPUs. What has AMZN’s stock done since your AWS acceleration call?
More like Big-Tech *gossip* for fintwit.
English
SemiAnalysis @dylan522p is the best semiconductor research team today. We're far from their level and have been learning from their research methodology.
Besides semiconductors, we spend more effort covering META, GOOG, AMZN, MSFT and dozens of software companies, so we're not really competing with SemiAnalysis in any way.
Dylan Patel@dylan522p
@SoftwareWatcher @GavinSBaker @fundaai @SemiAnalysis_ Huh? Gavin is one my longest running clients. We've worked together on a lot of things. He's not blocked. We even talked on the phone today.
English
@institLPGP @ShanuMathew93 @SemiAnalysis_ Yesterday's piece discusses near-term time-to-market solutions to power DCs when the grid can't keep up. In that context, solar is non-existent. Solar's role in energizing DCs is a separate topic.
English
English
@institLPGP @ShanuMathew93 @SemiAnalysis_ You are not understanding how Google and xAI's solar deployments work. They're completely different from the gas BTM deployments, and have a different purpose.
We'll cover that in a future piece. And we've explained it in detail to our Datacenter Model clients over a year ago.
English
@ShanuMathew93 @SemiAnalysis_ just wrote 5,000 words on btm power for dc… and didn’t mention solar even once. eve though it’s a core component of both xai and google strategies. illustrates the weakness of tourists in a space.
English
@puffinancial @GridGodAjey @dylan522p @SemiAnalysis_ Thanks for the flag, we had a few typos. However Lazard's analaysis has a few issues. First, it's not applicable to an islanded datacenter project, it doesn't capture all the required overbuild.
Second, their capex estimates are simply way too low.
English
ok @GridGodAjey @JeremieEO @dylan522p @SemiAnalysis_
what the hell is this? these LCOEs are not even close to reality. I’m not going to leak behind the paywall but you have a calculation error in your TCO. also please see the 2025 Lazard LCOE report to at least ballpark these. H-Class CCGT 100-200 🤯🤯🤯
English
Jeremie Eliahou Ontiveros รีทวีตแล้ว
ClusterMAX 2.0 is here!
In this update we rate 84 neoclouds, up from 26
Our market view tracks 209 providers, up from 169 in March and 124 last October
We spoke to over 140 end users as part of the research
And we wrote over 43,000 words about our experience
This is more than Animal Farm, but less than The Great Gatsby
More thoughts in 🧵
SemiAnalysis@SemiAnalysis_
ClusterMAX 2.0: The Industry Standard GPU Cloud Rating System 95% Coverage By Volume, 84 Providers Rated, 209 Providers Tracked, 140+ Customers Surveyed, 40,000 Words For Your Enjoyment newsletter.semianalysis.com/p/clustermax-2…
English
@RichardMeyerDC @ShanuMathew93 Depends on the type of chiller, but agree 1,000 for 1gw is a bit too high. Chiller specs are public, it's not hard to figure this out. You generally see ~1.5mw per air-cooled chiller and much more for water-cooled units.
We explained everything here: semianalysis.com/2025/02/13/dat…
English
@ShanuMathew93 1,000 industrial-scale chillers is not plausible for 1 GW. At 10GW that many chillers would require worst-case scenarios, high cooling loads, smaller inefficient machines, and virtually no economization.
English
Some stats at 1GW data center scale:
>1 GW = $50BN-$60BN (Jensen, NVIDIA)
>10 GW = 4-5M GPUs so ~500K GPUs per 1GW (Jensen, NVIDIA)
>% Cost of Build: (SemiAnalysis): Accelerators and servers: ~30% of build (GPUs + custom ASICs), Networking and optics: ~15%, Power and cooling: ~35%, Building and shell: ~20% (SemiAnalysis)
>Can require up to 1,000 industrial-scale chillers (Joe, Phaidra)
English
Jeremie Eliahou Ontiveros รีทวีตแล้ว
Too often, AI and energy conversations happen in separate rooms. That gap inspired @themeghamachine and I to launch the Watt & Why Series ⚡
Excited to kick this off and big thanks to @CrusoeAI and @SemiAnalysis_ for co-hosting the first session.
A space that brings everyone from AI researchers to energy traders together for the conversations that will shape our future!
ADM Investor Services@TradeADMIS
Come join the #ADMIS Houston Team Oct 16! Watt & Why: Energy x AI Talk in San Francisco 🌉 A special session on how hyperscale data centers reshaping power systems in our AI future. Seats are limited. ➡️ Request to join: hubs.ly/Q03H463k0
English
Amazon and Anthropic clearly aren't falling behind on GenAI :)
SemiAnalysis@SemiAnalysis_
Amazon’s AI Resurgence: AWS & Anthropic’s Multi-Gigawatt Trainium Expansion Anthropic multi-gigawatt clusters, Trainium ramp, best TCO per memory bandwidth, system-level roadmap, Bedrock and internal models semianalysis.com/2025/09/03/ama…
English
@BobEUnlimited Great piece Bob, but you're failing to account for the multiplier effects. AI infra is driving US Utility CapEx to increase $50b YoY in 2025 vs 2024. Datacenter construction spending too. And all the employment growth associated with this, as well as their consumer spending.
English
Can AI Alone Save the Economy?
These days bulls often say that AI-related investment and revenue is more than enough to keep the economy and markets powering higher. But any reasonable look at the numbers challenges these hopes. Link next.
English
@0xswizzcake @IamEmily2050 @SemiAnalysis_ How much AI revenue does Azure make excl. OpenAI? It's tiny.
What's the ARR of M365 Copilot and GH Copilot? It's decent, but disappointing given their head start. GH Copilot is losing share fast vs other coding agents.
English
Jeremie Eliahou Ontiveros รีทวีตแล้ว
Microsoft datacenter strategy – still in⏸️mode 🧵
@JeremieEO
Four months ago we flagged Microsoft’s pause on more than 1.5 GW of self-build capacity — on top of the leasing freeze we called out in December. Microsoft’s latest 10-K backs us up: committed datacenter construction is down year over year, almost no new leases were signed in the past 12 months, and land purchases have plunged relative to FY 24.
English
Jeremie Eliahou Ontiveros รีทวีตแล้ว
Robotics Levels of Autonomy
Single-Purpose Robots Automating Hundreds of Jobs
Pick And Place with low Autonomy is Expensive
General Purpose Autonomy Navigating
And Inspecting Large Sites
Targeting Low-Skill Labor In Early Pilots with Promise
Autonomy Capable Of Any Task in Research
semianalysis.com/2025/07/30/rob…
English
Jeremie Eliahou Ontiveros รีทวีตแล้ว
Last week, @mims asked us if NVIDIA’s claims for 30x inference performance with Blackwell were exaggerated.
We said yes, 30x improvement is gen-on-gen, doesn’t consider perf/$, and only applies in specific scenarios (large MOE’s).
So, Chris put me on the front page of wsj dot com saying that Groq can “deliver inference at a lower cost than is possible with NVIDIA systems”.
Is Groq’s technology credible? It’s an interesting question that I didn’t answer.
Lets unpack it.
English