AI models that cost $1B to train are underway, $100B models coming
AI training costs are rising exponentially, with models now reaching $1 billion to train. Companies are developing more powerful hardware to meet demand, but concerns about societal impact persist.
Read original article
AI training costs are skyrocketing, with models now costing up to $1 billion to train, according to Anthropic CEO Dario Amodei. Current models like ChatGPT-4o cost around $100 million, but Amodei predicts costs could reach $10 to $100 billion in the next few years. This exponential growth in costs is driven by the development of artificial general intelligence (AGI) from generative artificial intelligence. The hardware requirements for training these advanced models are also increasing, with estimates suggesting the need for more powerful chips and significant power consumption. Companies like Nvidia, AMD, and Intel are working to meet the demand for AI hardware, with projections indicating a surge in GPU data center deliveries. Concerns about the impact of AI advancements on society, including potential job displacement and energy consumption, are also being raised. As AI models become more powerful, questions about their economic viability and societal implications remain unanswered.
Related
Goldman Sachs says the return on investment for AI might be disappointing
Goldman Sachs warns of potential disappointment in over $1 trillion AI investments by tech firms. High costs, performance limitations, and uncertainties around future cost reductions pose challenges for AI adoption.
Gen AI: too much spend, too little benefit?
Tech giants and entities invest $1 trillion in generative AI technology, including data centers and chips. Despite substantial spending, tangible benefits remain uncertain, raising questions about future AI returns and economic implications.
Can the climate survive the insatiable energy demands of the AI arms race?
Google's emissions spike 50% in 5 years due to AI energy needs, posing climate challenges. Tech firms invest in renewables, but face infrastructure hurdles. AI advancements may paradoxically drive energy consumption.
AI industry needs to earn $600B per year to pay for hardware spend
The AI industry faces challenges to meet $600 billion annual revenue target to offset massive hardware investments. Concerns arise over profitability gap, urging companies to innovate and create value for sustainability.
Can the climate survive the insatiable energy demands of the AI arms race?
Google's emissions spike 50% in 5 years due to AI energy demands, posing climate challenges. Datacentres crucial for AI consume much electricity, with predictions of doubling consumption by 2026. Tech firms invest in renewables, but scaling challenges persist. AI's pursuit of advanced models intensifies energy use, raising sustainability concerns.
11 comments$100m is manageable, if you've got 100m paying subscribers or companies using your API for a year you can recoup the costs, but there aren't many companies with 100m users to monetise for it. $1B feels like it's pushing it, only a few companies in the world can monetise, and realistically it's about lasting through the next round to be able to continue competing, not about making the money back.
$100B though, that's a whole different game again. That's like asking for the biggest private investment ever made, for capex that depreciates at $50B a year. You'd have to be stupid to do it. The public markets wouldn't take it.
Investing that much in hardware that depreciates over 5+ years and is theoretically still usable at the end, maybe, but even then the biggest companies in the world are still spending an order of magnitude less per year, so the numbers end up working out very differently. Plus that's companies with 1B users ready to monetise.
This is the Anthropic CEO talking up his company's capital needs to the Norwegian Sovereign Wealth Fund ( Norges Bank Investment Management ) and trying to justify some absurd 100bn valuation.
Or is it the total overall cost of buying TPUs / GPUs, developing infrastructure, constructing data centers, putting together quality data sets, doing R&D, paying salaries, etc. as well as training the model itself? I could see that overall investment into AI scaling into the tens of billions over the next few years.
I wonder which timelines had this scenario…
Related
Goldman Sachs says the return on investment for AI might be disappointing
Goldman Sachs warns of potential disappointment in over $1 trillion AI investments by tech firms. High costs, performance limitations, and uncertainties around future cost reductions pose challenges for AI adoption.
Gen AI: too much spend, too little benefit?
Tech giants and entities invest $1 trillion in generative AI technology, including data centers and chips. Despite substantial spending, tangible benefits remain uncertain, raising questions about future AI returns and economic implications.
Can the climate survive the insatiable energy demands of the AI arms race?
Google's emissions spike 50% in 5 years due to AI energy needs, posing climate challenges. Tech firms invest in renewables, but face infrastructure hurdles. AI advancements may paradoxically drive energy consumption.
AI industry needs to earn $600B per year to pay for hardware spend
The AI industry faces challenges to meet $600 billion annual revenue target to offset massive hardware investments. Concerns arise over profitability gap, urging companies to innovate and create value for sustainability.
Can the climate survive the insatiable energy demands of the AI arms race?
Google's emissions spike 50% in 5 years due to AI energy demands, posing climate challenges. Datacentres crucial for AI consume much electricity, with predictions of doubling consumption by 2026. Tech firms invest in renewables, but scaling challenges persist. AI's pursuit of advanced models intensifies energy use, raising sustainability concerns.