Video yükleniyor...

Video Yüklenemedi

Ana Sayfaya Dön

Microsoft CEO Satya Nadella says computing power is now doubling every 6 months, as the Scaling Laws paradigm has taken over from Moore's Law, and the new currency is tokens per dollar per watt

tsarnick's profile picture

Tsarathustra

82,638 subscribers

553,390 görüntüleme • 1 yıl önce •via X (Twitter)

Haberler & PolitikaEğitimBilim & Teknoloji
LIVE

Anya Rossi• Live Now

Private livecam show

10 Yorum

Tsarathustra profil fotoğrafı
Tsarathustra1 yıl önce

Source:

Tim Ellis profil fotoğrafı
Tim Ellis1 yıl önce

So why does it still take 30 seconds for Word to open.

Rand profil fotoğrafı
Rand1 yıl önce

The end of Moore's law.. Moore's law squared is here

RNM profil fotoğrafı
RNM1 yıl önce

The metric “Tokens per dollar per watt” is a flawed metric that essentially refers to the efficiency of computing in terms of AI processing. Here’s my take: the most balanced metric would be tokens / second / watt-hour. It covers everything: computational throughput, energy efficiency, and hardware capability—all in one. As AI scales, this kind of precision is key for both tech and sustainability goals.

Compartk 𝕏/🚀🧬🗽🟨⬛ profil fotoğrafı
Compartk 𝕏/🚀🧬🗽🟨⬛1 yıl önce

Kurzweil was right since forty years ago

Leon Lang profil fotoğrafı
Leon Lang1 yıl önce

I didn’t get it. — he plots compute over time, which is not scaling law. — he says performance gets cheaper (which I understood to mean loss), but I don’t understand how performance is related to tokens per dollar per watt.

Kevin Brent Cook profil fotoğrafı
Kevin Brent Cook1 yıl önce

Jensen tried to tell us what was coming 8 years ago…

Dr. Bobby Gomez-Reino profil fotoğrafı
Dr. Bobby Gomez-Reino1 yıl önce

tokens per dollar per watt?? but....wtf. na. no words

₿lackthorne profil fotoğrafı
₿lackthorne1 yıl önce

What exactly does it mean to get an x amount of tokens per dollar, and then also per watt?

Hassan Abedi 📚🌿🦉 profil fotoğrafı
Hassan Abedi 📚🌿🦉1 yıl önce

It's over this time 👀

Benzer Videolar

Microsoft CEO Satya Nadella says the AI scaling laws multiply intelligence in both compute and algorithmic dimensions: for every 10x increase in compute, capability increases 100x
0:35

Microsoft CEO Satya Nadella says the AI scaling laws multiply intelligence in both compute and algorithmic dimensions: for every 10x increase in compute, capability increases 100x

Tsarathustra

80,676 görüntüleme • 1 yıl önce

Jensen Huang says AI Scaling Laws will increase computing power by a million times in the next decade, compared to Moore's Law which increased at 100 times per decade
0:46

Jensen Huang says AI Scaling Laws will increase computing power by a million times in the next decade, compared to Moore's Law which increased at 100 times per decade

Tsarathustra

140,852 görüntüleme • 1 yıl önce

$NVDA has the “lowest cost per token in the world” which is really a claim about owning the economics of inference. The company that delivers the most tokens per watt, per rack and per dollar is the one that captures the most demand and the best margins as AI scales.
0:47

$NVDA has the “lowest cost per token in the world” which is really a claim about owning the economics of inference. The company that delivers the most tokens per watt, per rack and per dollar is the one that captures the most demand and the best margins as AI scales.

Shay Boloor

48,594 görüntüleme • 4 ay önce

Microsoft CEO Satya Nadella says that "science itself is becoming computed" and increasing innovation will be driven by AI and quantum computers, leading to the industry becoming transformed
1:04

Microsoft CEO Satya Nadella says that "science itself is becoming computed" and increasing innovation will be driven by AI and quantum computers, leading to the industry becoming transformed

Tsarathustra

42,813 görüntüleme • 1 yıl önce

🎙️ What Is the Tau (τ) Law and How Does It Differ from Moore's Law? As Moore's Law approaches its limits, the semiconductor industry is exploring new ways to boost computing power. In this DeepTalk clip, we break down Huawei's proposed Tau (τ) Law and discuss how system architecture, advanced packaging, and chip interconnects may shape the future of AI computing. #DeepTalk #Semiconductors #AI #ChipDesign #MooresLaw
0:57

🎙️ What Is the Tau (τ) Law and How Does It Differ from Moore's Law? As Moore's Law approaches its limits, the semiconductor industry is exploring new ways to boost computing power. In this DeepTalk clip, we break down Huawei's proposed Tau (τ) Law and discuss how system architecture, advanced packaging, and chip interconnects may shape the future of AI computing. #DeepTalk #Semiconductors #AI #ChipDesign #MooresLaw

China Asset Management

10,669 görüntüleme • 2 ay önce

Microsoft CEO Satya Nadella's 3 Rules for Leadership in the AI Age "The magic of these three things - which is bringing clarity, creating energy, and driving success by solving over constrained problems - is what leadership is about." Satya Nadella talking with Garry Tan
1:18

Microsoft CEO Satya Nadella's 3 Rules for Leadership in the AI Age "The magic of these three things - which is bringing clarity, creating energy, and driving success by solving over constrained problems - is what leadership is about." Satya Nadella talking with Garry Tan

Overlap: Business & Tech

19,344 görüntüleme • 1 yıl önce

Here is the full conversation today between Microsoft CEO Satya Nadella and Elon Musk. Elon: "With Grok 3.5, which is about to be released, it's trying to reason from first principles."
6:43

Here is the full conversation today between Microsoft CEO Satya Nadella and Elon Musk. Elon: "With Grok 3.5, which is about to be released, it's trying to reason from first principles."

Sawyer Merritt

694,120 görüntüleme • 1 yıl önce

Dane Vahey of OpenAI says the cost per million tokens has fallen from $36 to $0.25 in the past 18 months, and as such AI is the greatest cost-depreciating technology ever invented
1:15

Dane Vahey of OpenAI says the cost per million tokens has fallen from $36 to $0.25 in the past 18 months, and as such AI is the greatest cost-depreciating technology ever invented

Tsarathustra

575,475 görüntüleme • 1 yıl önce

You can now build with Replit and deploy straight to Microsoft Fabric. 🙌 Code is the easy part. The backend is the hard part: identities, storage, schemas, governance. You build in Replit, and the app and data deploy into your enterprise-managed Microsoft Fabric tenant. MSFT CEO Satya Nadella announces in his Microsoft Build 2026 opening keynote.🔥
0:55

You can now build with Replit and deploy straight to Microsoft Fabric. 🙌 Code is the easy part. The backend is the hard part: identities, storage, schemas, governance. You build in Replit, and the app and data deploy into your enterprise-managed Microsoft Fabric tenant. MSFT CEO Satya Nadella announces in his Microsoft Build 2026 opening keynote.🔥

Manny Bernabe

15,431 görüntüleme • 2 ay önce

Microsoft CEO, Satya Nadella: "AI now generates 20–30% of Microsoft’s integrated code" the company is seeing mixed results in AI-generated code across different languages, with more progress in Python and less in C++
1:01

Microsoft CEO, Satya Nadella: "AI now generates 20–30% of Microsoft’s integrated code" the company is seeing mixed results in AI-generated code across different languages, with more progress in Python and less in C++

Haider.

141,788 görüntüleme • 1 yıl önce

Everyone’s obsessed with cheaper tokens. Meanwhile, enterprise AI spend keeps going up. Chamath says costs are doubling every 45 days for ~5% productivity gains. Glean’s founder says the metric that actually matters is cost per successful task, not cost per token.
1:09

Everyone’s obsessed with cheaper tokens. Meanwhile, enterprise AI spend keeps going up. Chamath says costs are doubling every 45 days for ~5% productivity gains. Glean’s founder says the metric that actually matters is cost per successful task, not cost per token.

Not Jerome Powell

45,215 görüntüleme • 27 gün önce

Artificial intelligence is fueling the American economy right now. But so much of the spending around AI is one massive bet -- and many on Wall Street are worried about whether the payout will materialize. I discussed with Microsoft CEO Satya Nadella in part 1 of our conversation:
4:10

Artificial intelligence is fueling the American economy right now. But so much of the spending around AI is one massive bet -- and many on Wall Street are worried about whether the payout will materialize. I discussed with Microsoft CEO Satya Nadella in part 1 of our conversation:

Fareed Zakaria

97,089 görüntüleme • 22 gün önce

At GTC, Nvidia CEO Jensen Huang highlighted a new equation: Revenue = Tokens per Watt × Total Available Gigawatts. Normal Computing CEO Faris Sbahi 🏴‍☠️ says this explains the industry’s push for energy-efficient and workload-specific chips: "Not that many years from now, rather than having a very small number of different chips running all of our workloads, we're going to have hundreds, if not thousands of different chips tailored to each workload and application in the data center."
1:20

At GTC, Nvidia CEO Jensen Huang highlighted a new equation: Revenue = Tokens per Watt × Total Available Gigawatts. Normal Computing CEO Faris Sbahi 🏴‍☠️ says this explains the industry’s push for energy-efficient and workload-specific chips: "Not that many years from now, rather than having a very small number of different chips running all of our workloads, we're going to have hundreds, if not thousands of different chips tailored to each workload and application in the data center."

TBPN

11,362 görüntüleme • 4 ay önce

$AMD CEO Lisa Su says the MI355 chip is the best inference accelerator in the market, delivering 40% more tokens per $ than any other chip. I think $AMD is a 10X-20X pick from here.
2:37

$AMD CEO Lisa Su says the MI355 chip is the best inference accelerator in the market, delivering 40% more tokens per $ than any other chip. I think $AMD is a 10X-20X pick from here.

Antonio Linares

174,643 görüntüleme • 1 yıl önce

AI has turned data centers from 5–10 kW racks into power-hungry AI factories. At that scale, the question isn’t “who has the most FLOPs?” It’s: Who can deliver the most intelligence per watt and per dollar? Nvidia’s GPUs still dominate, but inside Google there’s a different answer: the Tensor Processing Unit (TPU):
1:09

AI has turned data centers from 5–10 kW racks into power-hungry AI factories. At that scale, the question isn’t “who has the most FLOPs?” It’s: Who can deliver the most intelligence per watt and per dollar? Nvidia’s GPUs still dominate, but inside Google there’s a different answer: the Tensor Processing Unit (TPU):

Computer

31,075 görüntüleme • 8 ay önce

as requested, here is my compilation of every time natasha lyonne as nadia vulvokov says “fuck” in russian doll. the total fuck count is 286 over the course of 15 episodes, averaging 19 fucks from her mouth per episode and 1.4 fucks per minute. she is my hero ♥️
2:18

as requested, here is my compilation of every time natasha lyonne as nadia vulvokov says “fuck” in russian doll. the total fuck count is 286 over the course of 15 episodes, averaging 19 fucks from her mouth per episode and 1.4 fucks per minute. she is my hero ♥️

ellisima 🪆

10,428 görüntüleme • 2 yıl önce

#WATCH | Delhi: Union Minister of Health and Family Welfare JP Nadda says, "This is the first program that we are launching after coming to power...In 2014 we started 'Mission Indradhanush' and in 2024 we are scaling up the duration of the campaign from a fortnight to two months...In 2014, the mortality rate was 45 per 1000 live births which has been reduced to 32 per 1000 live births..."
4:36

#WATCH | Delhi: Union Minister of Health and Family Welfare JP Nadda says, "This is the first program that we are launching after coming to power...In 2014 we started 'Mission Indradhanush' and in 2024 we are scaling up the duration of the campaign from a fortnight to two months...In 2014, the mortality rate was 45 per 1000 live births which has been reduced to 32 per 1000 live births..."

ANI

14,630 görüntüleme • 2 yıl önce