bitcoin
Bitcoin (BTC) $ 66,574.84
ethereum
Ethereum (ETH) $ 3,206.27
tether
Tether (USDT) $ 1.00
bnb
BNB (BNB) $ 608.13
xrp
XRP (XRP) $ 0.545608
cardano
Cardano (ADA) $ 0.508373
usd-coin
USDC (USDC) $ 1.00
matic-network
Polygon (MATIC) $ 0.733759
binance-usd
BUSD (BUSD) $ 1.01
dogecoin
Dogecoin (DOGE) $ 0.160302
okb
OKB (OKB) $ 55.90
polkadot
Polkadot (DOT) $ 7.40
shiba-inu
Shiba Inu (SHIB) $ 0.000027
tron
TRON (TRX) $ 0.112602
uniswap
Uniswap (UNI) $ 8.07
wrapped-bitcoin
Wrapped Bitcoin (WBTC) $ 66,452.80
dai
Dai (DAI) $ 1.00
litecoin
Litecoin (LTC) $ 84.77
staked-ether
Lido Staked Ether (STETH) $ 3,204.48
solana
Solana (SOL) $ 155.46
avalanche-2
Avalanche (AVAX) $ 38.68
chainlink
Chainlink (LINK) $ 15.35
cosmos
Cosmos Hub (ATOM) $ 8.79
the-open-network
Toncoin (TON) $ 5.65
ethereum-classic
Ethereum Classic (ETC) $ 28.08
leo-token
LEO Token (LEO) $ 5.75
filecoin
Filecoin (FIL) $ 6.52
bitcoin-cash
Bitcoin Cash (BCH) $ 509.94
monero
Monero (XMR) $ 123.64
bitcoin
Bitcoin (BTC) $ 66,574.84
ethereum
Ethereum (ETH) $ 3,206.27
tether
Tether (USDT) $ 1.00
bnb
BNB (BNB) $ 608.13
usd-coin
USDC (USDC) $ 1.00
xrp
XRP (XRP) $ 0.545608
binance-usd
BUSD (BUSD) $ 1.01
dogecoin
Dogecoin (DOGE) $ 0.160302
cardano
Cardano (ADA) $ 0.508373
solana
Solana (SOL) $ 155.46
matic-network
Polygon (MATIC) $ 0.733759
polkadot
Polkadot (DOT) $ 7.40
tron
TRON (TRX) $ 0.112602

Crypto AI Initiatives Would Have to Purchase Chips Value Their Whole Market Cap to Meet Ambitions

Must Read

The potential of text-to-video era excites the crypto market, and AI tokens rose when OpenAI first unveiled a demo of Sora

However to make this mainstream the compute energy shall be staggering. Extra server-grade H100 GPUs shall be required than Nvidia produces in a yr, or what its largest prospects run of their information facilities collectively.

What number of Graphics Processing Models (GPUs) shall be required to make text-to-video era mainstream? Lots of of 1000’s – and greater than are presently in use by Microsoft {{MSFT}}, Meta {{META}}, and Google {{GOOG}} mixed.

The primary demo of OpenAI’s text-to-video generator Sora amazed the world, and this renewed curiosity in Synthetic Intelligence (AI) tokens, with many surging within the aftermath of the demo.

Within the weeks that adopted, many crypto AI tasks emerged, additionally promising to do text-to-video and text-to-image era, and the AI token class now has a $25 billion market cap in line with CoinGecko information.

Behind the promise of AI-generated movies are armies of Graphics Processing Models (GPUs), the processors from the likes of Nvidia {{NVDA}} and AMD {{AMD}}, which make the AI revolution doable due to their potential to compute massive volumes of information.

However simply what number of GPUs will it take to make AI-generated video a mainstream factor? Greater than main massive tech firms had of their arsenal in 2023.

An Military of 720,000 Nvidia H100 GPUs

A latest analysis report by Factorial Funds estimates that 720,000 high-end Nvidia H100 GPUs are required to assist the creator neighborhood of TikTok and YouTube.

See also  Wormhole token slides 24% earlier than recovering barely following airdrop

Sora, Factorial Funds writes, requires as much as 10,500 highly effective GPUs for a month to coach, and might generate solely about 5 minutes of video per hour per GPU for inference.

Because the chart above demonstrates, coaching this requires considerably extra compute energy than GPT4 or nonetheless picture era.

With widespread adoption, inference will surpass coaching in compute utilization. Which means as extra individuals and corporations begin utilizing AI fashions like Sora to generate movies, the pc energy wanted to create new movies (inference) will turn into larger than the ability wanted to coach the AI mannequin initially.

To place issues in perspective, Nvidia shipped 550,000 of the H100 GPUs in 2023.

Information from Statista exhibits that the twelve largest prospects utilizing Nvidia’s H100 GPUs collectively have 650,000 of the playing cards, and the 2 largest—Meta and Microsoft—have 300,000 between them.

Assuming a price of $30,000 per card, it will take $21.6 billion to convey Sora’s desires of AI-generated text-to-video mainstream, which is sort of the whole market cap of AI tokens for the time being.

That is in case you can bodily purchase all of the GPUs to do it.

Nvidia is not the one sport on the town

Whereas Nvidia is synonymous with the AI revolution, it is vital to keep in mind that it isn’t the one sport on the town.

Its perennial chip rival AMD makes competing merchandise, and buyers have additionally handsomely rewarded the corporate, pushing its inventory from the $2 vary within the fall of 2012 to over $175 right now.

See also  dYdX group to vote on staking 20M DYDX with Stride

There are additionally different methods to outsource computing energy to GPU farms. Render (RNDR) gives distributed GPU computing, as does Akash Community (AKT). However the majority of the GPUs on these networks are retail-grade gaming GPUs that are considerably much less highly effective than Nvidia’s server-grade H100 or AMD’s competitors.

Regardless, the promise of text-to-video, which Sora and different protocols promise, goes to require a herculean {hardware} raise. Whereas it is an intriguing premise and will revolutionize Hollywood’s artistic workflow, do not count on it to turn into mainstream anytime quickly.

We will want extra chips.

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Latest Articles

What Occurs When the Final Bitcoin Is Mined?

Think about a world the place the cash provide is finite, and there’s no {dollars} left to print. Sounds...

More Articles Like This