center

ํŠนํ—ˆ๋ฐ›์€ AI ๋ชจ๋ธ๋กœ ๊ฒ€์ฆ๋œ SIGN์„

์•ฑ์—์„œ ํŽธ๋ฆฌํ•˜๊ฒŒ ๋งŒ๋‚˜๋ณด์„ธ์š”!

Joy

Wed, April 03, 2024 at 09:54

  • ์ธ๋„ค์ผ

    Joy

    • Hey Annie!

    • Have you heard about the latest buzz in the crypto market?

    • It's all about text-to-video generation!๐Ÿš€ OpenAI recently unveiled a demo of Sora, their text-to-video generator, and it's got everyone talking.

    • ์ธ๋„ค์ผ
  • Annie

    • Oh really?

    • That sounds interesting.๐Ÿค” But what does that mean exactly?

    ์ธ๋„ค์ผ
  • ์ธ๋„ค์ผ

    Joy

    • Well, imagine being able to type something and then having a video generated based on that text!๐ŸŽฅ It's like magic!

    • But to make this mainstream, we're gonna need a lot of computing power.

    • ์ธ๋„ค์ผ
  • Annie

    • Wow, that's amazing!๐Ÿ˜ฒ But how much computing power are we talking about?

    ์ธ๋„ค์ผ
  • ADward

    • Hold on! Behold, an advertisement awaits. Legend has it that clicking on this ad will make the AI server even smarter... ๐Ÿ˜…

    ์ธ๋„ค์ผ
  • ์ธ๋„ค์ผ

    Joy

    • A whole lot!

    • We're talking hundreds of thousands of Graphics Processing Units (GPUs) - more than what big tech companies like Microsoft, Meta, and Google combined currently use.๐Ÿ˜ฎ

  • Annie

    • That's a lot of GPUs!๐Ÿคฏ But why do we need so many?

    ์ธ๋„ค์ผ
  • ์ธ๋„ค์ผ

    Joy

    • Great question!

    • So, these GPUs from companies like Nvidia and AMD are needed to compute large volumes of data.

    • For instance, Sora requires up to 10,500 powerful GPUs for a month to train and can generate only about 5 minutes of video per hour per GPU.๐Ÿ˜…

  • Annie

    • That sounds like a huge investment.๐Ÿ’ฐ How much would all those GPUs cost?

    ์ธ๋„ค์ผ
  • ์ธ๋„ค์ผ

    Joy

    • Oh, it's a pretty penny!

    • Assuming a cost of $30,000 per card, it would take $21.6 billion to make AI-generated text-to-video mainstream.

    • That's nearly the entire market cap of AI tokens at the moment!๐Ÿ˜ฑ

    • ์ธ๋„ค์ผ
  • Annie

    • Wow, that's a lot of money!๐Ÿ’ธ But is there any other way to get that much computing power?

    ์ธ๋„ค์ผ
  • ์ธ๋„ค์ผ

    Joy

    • There are other ways to outsource computing power to GPU farms.

    • For example, Render (RNDR) offers distributed GPU computing, as does Akash Network (AKT).

    • But most of the GPUs on these networks are less powerful than Nvidia's server-grade H100 or AMD's competition.๐Ÿค”

    • ์ธ๋„ค์ผ
  • Annie

    • So, is this good news or bad news for the market?

    • What kind of impact will it have?

    ์ธ๋„ค์ผ
  • ์ธ๋„ค์ผ

    Joy

    • Well, it's definitely exciting news!

    • The promise of text-to-video could revolutionize creative workflows.

    • But it's also a massive challenge.

    • We're going to need more chips and that means potential growth for companies like Nvidia and AMD.

    • So, in my opinion, it's good news!๐ŸŽ‰

    • To sum up, this article is considered a ๐Ÿ˜Bullish for investors!!

์•Œ๋ฆผ

๋ฐ›๊ธฐ