Workflow
Compute Power
icon
Search documents
OpenAI co-founder on new deal with AMD: We need as much compute power as we can possibly get
CNBC Television· 2025-10-06 15:19
Shares of AMD are surging. Do I just use that word surging. How about like colossal gain.A company announced a massive new 5-year deal with Open AI. Cannot explain to people how big this is. Joining us now first on CNBC CEO Lisa Sue and Open AI's co-founder and president Greg Brockman.Uh congrat congratulations to both. Lisa, I'd like to start with you. Okay.Yeah. Good morning. Great to see you guys. Great to see you.Now Lisa, this is a huge coup uh because I think a lot of people felt that you were no near ...
Governments Pour Billions Into Chips And AI Infrastructure To Fuel Arms Race
Forbes· 2025-09-22 18:03
Group 1: Importance of Compute Power - Compute power is becoming a critical resource in 2025, akin to oil refineries and power plants in the 21st century [2] - Governments worldwide are investing billions in semiconductors and cloud infrastructure to gain economic advantages and lead in artificial intelligence (AI) [3] Group 2: Major Investments and Contracts - OpenAI signed a historic cloud contract with Oracle worth $300 billion over five years, requiring 4.5 gigawatts of power capacity [4] - Since the launch of ChatGPT in late 2022, the market capitalizations of major tech companies like Nvidia, Microsoft, Oracle, and Broadcom have increased by $8 trillion [5] - Construction spending on U.S. data centers reached an all-time high of $40 billion in June, a 30% increase from the previous year, following a 50% surge in 2024 [5] Group 3: Government Support and Strategic Assets - The U.S. government is actively supporting semiconductor companies, exemplified by Intel receiving an $8.9 billion equity stake and additional CHIPS Act grants [7] - The UK is also increasing AI investments, with Microsoft planning to invest $30 billion by 2028 to build a supercomputer, contributing to over $40 billion in total commitments from tech giants [9] Group 4: Industry Dynamics and Energy Demand - Nvidia is projected to lead the AI semiconductor market, with revenue expected to quadruple in the coming years, while Broadcom and AMD are also competing for market share [10] - The Department of Energy forecasts that AI-specific electricity use will grow by 33% annually, highlighting the need for sufficient energy supply to support the AI ecosystem [11] Group 5: Defense Spending and High-Tech Applications - Global defense budgets are rapidly increasing, with military spending projected to rise from just under $3 trillion in 2024 to potentially $4 trillion by 2030, focusing on high-tech systems [12] - The connection between semiconductors and defense is strengthening, as more military budgets are allocated to advanced technologies [12] Group 6: Labor Market and Automation - The labor market is showing signs of softening, with declining job openings and reduced wage hikes, coinciding with increased investment in AI [13][14] - The paradox of AI being marketed as a productivity booster during a moderating economy suggests a trend towards automation amid labor demand weakness [14] Group 7: Future Outlook - Compute power is viewed as the new energy, with nations stockpiling it and companies monetizing it, indicating its significance in defining the 21st century [15]
OpenAI, Oracle Deal Shows Need for Compute Power
Bloomberg Technology· 2025-07-03 17:56
AI and Computing Power Demand - The demand for computing power is insatiable, indicating a long runway for investment in physical infrastructure like semiconductors and cloud [1][2][3] - The expansion between Oracle and Open AI is 45 gigawatts worth of capacity, highlighting the significant energy requirements of AI development [3] - The U S is likely to lead the AI race due to its self-sufficiency and relatively cheap energy production compared to Europe and the U K [5] Energy Sector and Infrastructure - The energy sector and energy infrastructure requirements are intrinsically linked to the need for computing power, leading hyperscalers to invest in energy sources like nuclear [4] - The quest for cheaper, more efficient energy is crucial for supporting data center expansion and AI development [5] Market and Economic Factors - The jobs data is considered "Goldilocks" because of better-than-expected job numbers and a solid labor market, alongside slower wage growth [6][7] - The Federal Reserve is in no urgency to cut rates, which the market seems to prefer, valuing a strong labor market over immediate rate cuts [7] Technology Sector Performance - Technology sector exhibits strong earnings growth and benefits from productivity gains, making it an attractive area for investment [9] - Lower bond yields and interest rates, potentially resulting from future rate cuts, would particularly benefit growth areas like the technology sector [10]