Simone

687 posts

Simone banner
Simone

Simone

@Simne1core

🇪🇺 🇮🇹 Semiconductor | AI Infrastructure | Energy & Supply Chain

Italy Katılım Şubat 2026
215 Takip Edilen61 Takipçiler
Simone
Simone@Simne1core·
Questo è il punto che l’Europa dovrebbe studiare. Huawei tratta i vincoli semiconductor come un problema industriale da aggirare. L’Europa troppo spesso li tratta come un tema da policy paper. La sovranità tecnologica non si dichiara, si fabbrica. #Huawei #EU #Semiconductor #OpenSemiconductorRadar
Huawei@Huawei

HUAWEI's He Tingbo has presented the Tau (τ) Scaling Law, a new principle for guiding the future development of the semiconductor industry, highlighting why time scaling can deliver strong benefits across device, circuit, chip, and system.

Italiano
0
0
0
32
Simone
Simone@Simne1core·
@OrevaZSN La distribuzione culturale ha altri vincoli: gusto, timing, fiducia, community e contesto. L’AI può produrre contenuto infinito, ma non significa che quel contenuto abbia valore sociale.
Italiano
0
0
0
5
𐌁𐌉Ᏽ 𐌕𐌉𐌌𐌉
Teenagers have started calling AI art "boomer art" and consider it cringe, and YouTubers have stopped using AI-generated thumbnails because teenagers find them cringe and won't click on them. I honestly couldn't be happier.
English
441
10.8K
105K
930.5K
Simone
Simone@Simne1core·
@Ric_RTP L’AI non sostituisce automaticamente il lavoro umano. Prima deve battere il costo totale del sistema: compute, energia, tool, integrazione, errori e supervisione. Il software sembra magico. La bolletta del datacenter è molto reale.
Italiano
0
0
0
36
Ricardo
Ricardo@Ric_RTP·
Microsoft just banned its own engineers from using AI. The tool was literally costing MORE than the humans it was supposed to replace. They lied to you about AI adoption and now the whole narrative is blowing up: Microsoft gave thousands of engineers access to Claude Code six months ago and encouraged them to use it. Engineers loved it and adoption exploded. But then the invoices arrived. Token-based pricing means every query, every code review, every debugging session costs money. At scale across 100,000 engineers, the numbers became so large that Microsoft issued an internal order to cancel nearly all Claude Code licenses by end of June and force everyone onto their own cheaper tool instead. The company that invested $5 billion in Anthropic just told its own people to stop using Anthropic's product because it costs too much. Uber's story is even worse... Their CTO Praveen Neppalli Naga told The Information that the budget he planned for the full year was "blown away already" by April. Uber had rolled out Claude Code in December 2025. By March, 84% of their 5,000 engineers were using it with 70% of all committed code coming from AI systems. Heavy users were burning $500 to $2,000 per month each. Naga himself spent $1,200 in a single two-hour demo session. The company had even built internal leaderboards ranking engineers by how much AI they used. They literally gamified the spending and then ran out of money. Now look at what Nvidia's own VP of applied deep learning Bryan Catanzaro said to Axios last month. Direct quote: "For my team, the cost of compute is far beyond the costs of the employees." This is a VP at the company that SELLS the chips saying that using AI is more expensive than paying humans. Think about what this means for the entire AI narrative. Every CEO on every earnings call for the past two years has said the same thing: AI will make us more efficient, reduce headcount, and cut costs. The stock market rewarded every company that said it. Fired workers, stock goes up. Announced AI adoption, stock goes up. But the actual companies deploying AI at scale are discovering the math doesn't work. The MORE employees use AI, the HIGHER the bill. Goldman Sachs forecasts a 24x increase in token consumption by 2030 as companies adopt AI agents. Gartner just published a report showing that even though individual token prices will drop 90% by 2030, total enterprise AI costs will go UP because agents consume exponentially more tokens per task than basic tools. Meta built an internal dashboard called "Claudeonomics" to track which employees use the most AI. Amazon started pushing engineers to "tokenmaxx," their internal term for consuming as many AI tokens as possible. Both companies are spending hundreds of billions on AI infrastructure this year alone. And Microsoft, the company that bet its entire future on AI, just told 100,000 engineers to stop using the tool they liked best because the per-token bills got out of control. The companies building AI are telling investors it saves money. The companies using AI are finding out it costs more than the humans it was supposed to replace. And even the company that makes the chips just admitted it through its own VP. This is the gap nobody on Wall Street is pricing in. $725 billion in AI infrastructure spending this year across Big Tech. And the first companies to actually deploy these tools at scale are already pulling back because the economics don't work. What do you think?
English
1.7K
8.5K
20.9K
2.3M
Simone
Simone@Simne1core·
@business @norveclifinance Everyone wants the TSMC shortcut. But semiconductor independence does not come from slogans. It comes from process trade-offs, equipment limits, packaging choices, yield pain and years of manufacturing discipline.
English
0
0
1
309
Bloomberg
Bloomberg@business·
Huawei says it has come up with a new pathway to shorten its gap with industry leader TSMC, potentially achieving a breakthrough in making advanced semiconductors without cutting-edge equipment bloomberg.com/news/articles/…
English
58
159
686
267.2K
Simone
Simone@Simne1core·
HBM e wafer capacity sono i colli di bottiglia visibili. Ma sotto i GPU, CPU, ASIC e networking chip c’è un layer meno sexy: ABF substrate. Senza substrate ad alte prestazioni, il chip non diventa sistema. L’AI non dipende solo dal die. Dipende dal package. #HBM #Semiconductors #SupplyChain
Simone tweet media
Italiano
0
0
0
22
Simone
Simone@Simne1core·
@NuttyCLD At these power densities, passives are no longer “background components”. Silicon capacitors, voltage regulation and package-level power stability become part of compute performance. AI is turning small components into strategic capacity.
English
0
0
0
232
Nutty
Nutty@NuttyCLD·
Samsung Electro-Mechanics and ADI Enter Silicon Capacitors On May 20, Samsung Electro-Mechanics signed a ~$1.1 billion silicon capacitor supply contract with a U.S. big tech, running through end-2028. One day earlier, ADI announced a $1.5 billion acquisition of Empower Semiconductor. Same component. Two days. Silicon capacitors sit inside the AI chip package, right next to the GPU, stabilizing power at the closest possible point. The broader silicon cap market is around $4B and spread across multiple suppliers. But narrow it to high-end AI chip-packaging applications, and the market was effectively a Murata–TSMC duopoly. Two new entrants walked into that tightly held space. Samsung Electro-Mechanics extended its MLCC and FC-BGA substrate expertise into silicon caps. ADI absorbed Empower's portfolio into its power-management platform. Four different paths now lead to the same component: a passives maker, a foundry, a substrate maker, and an analog IC company. Silicon capacitors used to be a small niche market. AI's expansion is enlarging that niche, and incumbents are rising alongside the new entrants. That is reason enough to keep watching the component layer.
Nutty@NuttyCLD

x.com/i/article/2054…

English
4
20
167
26.7K
Simone
Simone@Simne1core·
@dykeseyah Il problema non è “AI buona” o “AI cattiva”. Il punto è molto più concreto: chi controlla compute, dati, energia e distribuzione controlla la capacità di produrre contenuto, decisioni e automazione su scala.
Italiano
0
0
0
23
Simone
Simone@Simne1core·
@elonmusk @SpaceX non sta solo lanciando hardware. Sta controllando il sistema end-to-end👏
Italiano
0
1
2
20
Simone
Simone@Simne1core·
@StockMKTNewz Nel ciclo AI server, compliance e supply chain non sono funzioni secondarie. Sono parte del prodotto. Se non sai controllare dove finisce l’hardware, prima o poi il problema non è più commerciale. È industriale e geopolitico.
Italiano
1
0
1
21
Evan
Evan@StockMKTNewz·
Nvidia CEO Jensen Huang urged Super Micro $SMCI to tighten up on compliance after Taiwan detained three people for allegedly making fraudulent declarations about AI servers - Bloomberg
Evan tweet media
English
42
13
331
39K
Simone
Simone@Simne1core·
@jwt0625 Sì. Il compute AI sta diventando un problema di system engineering. Chip, power delivery, interconnect e thermal design non sono più layer separati, sono lo stesso sistema fisico.
Italiano
0
0
0
29
outside five sigma
outside five sigma@jwt0625·
future computers will be heat exchangers with compute, communication, and power embedded in it
English
10
4
64
2.8K
Simone
Simone@Simne1core·
Questa non è una notizia sui prezzi, è una notizia sulla pianificazione industriale. Quando HBM/DRAM diventano scarse, non stai più ottimizzando il modello, stai competendo per capacità produttiva. E se la memoria diventa il prossimo vincolo strutturale, chi avrà più leverage nella supply chain AI: GPU vendors, memory makers o hyperscaler con contratti di lungo periodo?
Italiano
0
0
0
26
Pirat_Nation 🔴
Pirat_Nation 🔴@Pirat_Nation·
NVIDIA CFO Colette Kress directly addressed the ongoing RAM shortages and rising memory prices, blaming other companies for failing to prepare early enough. “Many of the companies, for example, are sitting here going, ‘Oh my gosh, the memory price went up.’ We knew that was going to happen. That was something everybody should have ordered a long time ago, at least we did.” According to Kress, NVIDIA secured its memory supply well before the AI-driven demand surge in October and November 2025. NVIDIA’s upcoming Rubin AI platform is expected to require as much as 6 billion GB of LPDDR memory by 2027, potentially exceeding the combined memory demand of both Apple and Samsung.
Pirat_Nation 🔴 tweet mediaPirat_Nation 🔴 tweet media
English
98
109
1.5K
124.6K
Simone
Simone@Simne1core·
Tutti parlano di AI come se fosse una corsa tra modelli. Ma sotto i modelli c’è una macchina industriale enorme: chip, HBM, packaging, power delivery, datacenter, energia, acqua, cooling, grid, equipment e supply chain. L’AI non è immateriale. È fisica. #AIInfrastructure #Semiconductors
Simone tweet media
Italiano
0
0
0
31
Simone
Simone@Simne1core·
Sarebbe stata una conversazione interessante. Il tema hardware è ancora troppo assente dal dibattito startup italiano. Si parla di modelli, di prodotto, di go-to-market. Quasi mai di cosa ci gira sotto. Quello che fai con Paradigma è esattamente il tipo di AI che ha bisogno di compute affidabile e continuo, ricerca autonoma, elaborazione di grandi volumi di dati scientifici. Ogni sistema del genere gira su silicio che non controlliamo. Prima o poi diventa il limite vero alla scalabilità. Se vuoi approfondire, sono qui.
Italiano
0
0
1
38
tensorqt
tensorqt@tensorqt·
@Simne1core agree, avrei voluto l’occasione di poter parlarne di più
Italiano
1
0
2
225
Simone
Simone@Simne1core·
Al Future Proof Society di Brescia. Sul palco Francesco Pappone parla di startup. La cosa che mi colpisce di più in questi eventi: tutti parlano di AI, tutti parlano di futuro. Ogni startup AI che nasce in Italia ha bisogno di compute. Quel compute gira su chip. Quei chip vengono da Taiwan. Costruire un ecosistema startup in Europa senza affrontare il problema dell'infrastruttura hardware è come costruire una Formula 1 senza pensare al carburante. Il talento c'è, le idee ci sono, ma manca ancora il layer fisico che le sostiene. #FutureProofSociety #Brescia #Startup #AIEuropa #Semiconductors
Simone tweet media
Brescia, Lombardy 🇮🇹 Italiano
1
0
1
319
Simone
Simone@Simne1core·
Bull è un caso interessante di sovranità industriale sul serio. Non un annuncio, non una roadmap, una fabbrica reale ad Angers che costruisce supercalcolatori dal 1931. La Francia ha capito una cosa che molti altri paesi europei non hanno ancora capito: la sovranità digitale non si compra, si costruisce nel tempo, con investimento paziente e continuità industriale. Il problema rimane però lo stesso di sempre, Bull assembla supercalcolatori, ma i chip dentro quei supercalcolatori vengono da dove? TSMC, Samsung, Nvidia... Finché l'Europa non controlla il silicio, controlla solo il contenitore, non il contenuto. Rispetto per Bull, ma la vera partita è un livello più in basso.
Italiano
0
0
0
16
Alex Xplore
Alex Xplore@AlexXplore·
🇫🇷 Fondée en 1931 en France, l’entreprise Bull est l’un des plus anciens acteurs européens de l’informatique. 🖥️ Spécialisée dans le calcul haute performance, les supercalculateurs et l’intelligence artificielle, elle dispose de la seule usine européenne de ce type à Angers. 🔄 Après une histoire mouvementée marquée par des nationalisations et privatisations, Bull a été rachetée par l’État français en mars 2026 pour renforcer la souveraineté numérique. 👨‍💻 Avec environ 3 000 salariés et un plan de 500 embauches en 2026, l’entreprise équipe les projets stratégiques européens en recherche, défense et simulation nucléaire. 💪 Symbole de l’ambition technologique française, Bull incarne aujourd’hui plus de 90 ans d’histoire au service de l’indépendance numérique de la France.
Alex Xplore tweet mediaAlex Xplore tweet media
Français
7
55
234
7.5K
Simone
Simone@Simne1core·
€400 milioni per il riscaldamento pulito. Benvenuti! Ma mettiamoli in contesto. Cinque hyperscaler americani spenderanno $700 miliardi in infrastruttura AI solo quest'anno. La risposta energetica europea si misura in centinaia di milioni, la decarbonizzazione del riscaldamento è importante, ma l'emergenza energetica reale in Europa adesso è un'altra. Amsterdam, Francoforte e Dublino hanno già saturato il loro grid. Diversi stati membri hanno vietato nuovi datacenter. L'Italia ha appena approvato 300 MW di capacità AI in Lombardia, senza ancora avere un piano energetico all'altezza. €400 milioni per il calore pulito non alimentano un singolo campus AI da 300 MW. L'Europa ha bisogno di entrambe le cose, energia pulita e scala. Adesso non sta facendo né l'una né l'altra abbastanza velocemente.
Italiano
0
0
0
15
European Commission
European Commission@EU_Commission·
We are accelerating the deployment of innovative clean heat technologies across European industry. These projects will help contribute to ✅the EU's clean transition ✅energy independence and security ✅industrial competitiveness link.europa.eu/MwyNjb
European Commission tweet media
English
19
20
89
7.9K
Simone
Simone@Simne1core·
Kioxia sta facendo la mossa più intelligente del momento, essere contrarian. Samsung e SK Hynix hanno ritardato la NAND di 10a generazione perché la difficoltà tecnica è reale. Impilare 430 layer spinge al limite i processi di deposizione e incisione. Ma quel ritardo crea una finestra. Se Kioxia esegue mentre i coreani esitano, non chiude solo il gap, ribalta la narrativa da "eterno terzo posto" a "leader tecnologico su un nodo specifico." È esattamente come Micron ha recuperato su HBM non essendo il migliore in assoluto, ma eseguendo su una generazione specifica mentre i rivali erano distratti. La gara NAND è appena diventata molto più interessante.
Italiano
0
0
3
454
Jukan
Jukan@jukan05·
Very interesting. With Samsung and SK Hynix delaying V10 NAND mass production to next year, if Kioxia successfully ramps V10 NAND next year, the technology gap between Samsung/SK Hynix and Kioxia would narrow significantly.
Jukan tweet media
Jukan@jukan05

Kioxia Signals Intent to Mass-Produce 10th-Gen NAND Next Year, Chasing Samsung and SK Hynix Japan's Kioxia has flagged next-generation NAND mass production as a key priority for next year. With Korean players such as Samsung Electronics and SK Hynix pushing back the commercialization timeline for their own next-gen NAND, the move is read as a strategy to rapidly close the technology gap. According to industry sources on the 23rd, Kioxia is planning to begin mass production of 10th-generation (BiCS 10) NAND next year. Kioxia is one of Japan's major NAND manufacturers. Per market research firm TrendForce, the company held a 14.1% share of the global NAND market in the fourth quarter of last year, ranking third—behind Samsung Electronics (28.0%) and SK Hynix (22.1%). There is also a view that Kioxia is ramping up its technology at pace. In its most recent earnings release, the company identified "the launch of 10th-generation BiCS NAND" as one of its strategic priorities for fiscal year 2026 (April 2026–March 2027). Its capital expenditure in the latest quarter was likewise concentrated on 8th- and 10th-generation NAND. NAND has evolved by vertically stacking ever more cells—the smallest unit of memory storage. Kioxia applies its proprietary cell-stacking technology, BiCS (Bit Cost Scalable). The 10th-generation BiCS stacks a total of 332 layers, increasing storage capacity per unit area by 59% and improving data transfer speed by 33% versus the prior generation (218 layers). That said, just how much 10th-gen NAND Kioxia will actually mass-produce remains unclear. One semiconductor industry official noted, "Kioxia had originally planned to begin investing in 10th-gen NAND as early as the second half of last year, but there are still no confirmed orders," adding that "a clearer picture should emerge around the second half of this year." Should Kioxia indeed move forward with mass-production investment in 10th-gen NAND, it is expected to be able to narrow the technology gap with Samsung Electronics and SK Hynix more quickly. Korean firms are also developing 10th-gen NAND, but have yet to execute actual mass-production investment. Samsung Electronics had originally planned to mass-produce its 430-layer-class 10th-gen NAND this year, but the plan has been pushed back to at least next year. The high technical difficulty of realizing 10th-gen NAND, along with market demand, are seen as contributing factors. A source familiar with the matter said, "Samsung Electronics is deliberating over the timing of its 10th-gen NAND conversion investment and has not yet shared any concrete equipment-ordering plans," adding that "the situation is the same for SK Hynix."

English
14
98
702
197.9K
Simone
Simone@Simne1core·
At Future Proof Society in Brescia today. The room is packed, everyone is talking about AI, energy, infrastructure. The same conversations I have every day on here, except in person. What strikes me most: the urgency is real. People are starting to understand that the AI race isn't won with software, it's won with chips, power, and physical infrastructure. The gap between what Europe is discussing and what it's actually building is still too wide. But at least the conversation is happening. That's how it starts. #FutureProofSociety #FPS2026 #Brescia #AIInfrastructure #Semiconductors #EnergyInfrastructure #EuropeAI #TechItalia #Italia #Innovation #Renzi
Simone tweet media
Brescia, Lombardy 🇮🇹 English
0
0
0
58
Simone
Simone@Simne1core·
Ottime notizie per Milano, ma poniamoci la domanda più difficile. Anthropic apre un ufficio a Milano, Claude utilizza GPU Nvidia, queste GPU sono prodotte da TSMC a Taiwan, stampate da macchine ASML nei Paesi Bassi. Alimentato da energia che l'Italia importa in gran parte. I talenti nel campo dell'IA stanno arrivando in Italia. L'hardware che lo fa funzionare è ancora in parte altrove, un ufficio è un buon inizio.
Seb Johnson@SebJohnsonUK

Anthropic is doubling down on Europe and opening an office in Milan. Europe is Anthropic's fastest growing region with revenue up 9x YoY, while Milan is the centre of Italian tech. Anthropic now has hubs in London, Dublin, Zurich, Paris, Munich and Milan. It's pushing more and more into Europe as its relationship with the US continues to sour. Great stuff for Italy!

Italiano
0
0
0
106