OpenAI, Oracle Expand Stargate Project | Bloomberg Tech 7/3/2025

Key Concepts

• Compute Capacity: The amount of processing power available, particularly for AI and cloud computing.
• Stargate Initiative: OpenAI's project to expand its computing infrastructure.
• Gigawatt (GW): A unit of power, often used to measure the capacity of power plants or data centers.
• Cloud Service Provider (CSP): A company that provides cloud computing services, such as Oracle.
• Capital Expenditure (CapEx): Spending on fixed assets, such as data centers and equipment.
• Graphics Processing Unit (GPU): A specialized electronic circuit designed to rapidly manipulate and alter memory to accelerate the creation of images in a frame buffer intended for output to a display device.
• Advanced Micro Devices (AMD): A semiconductor company that designs and manufactures computer processors and graphics cards.
• Hyperscalers: Companies with massive computing infrastructure, such as Amazon, Google, and Microsoft.
• Electronic Design Automation (EDA): Software used to design and manufacture semiconductors.
• Tariffs: Taxes imposed on imported goods or services.
• Private Equity (PE): Investment in companies that are not publicly traded.
• Seed Investing: Providing initial capital to early-stage startups.
• AI Native: Companies that are built from the ground up using artificial intelligence.
• LLM's: Large Language Models.

OpenAI and Oracle's Expanded Stargate Project

• Deal Details: OpenAI is expanding its "Stargate" project with Oracle, securing 4.5 gigawatts of compute capacity. This is potentially the largest cloud deal ever, involving tens of billions of dollars in chips and power infrastructure.
• Scale: 4.5 gigawatts is equivalent to the power output of 4.5 nuclear reactors, enough to power approximately 3.5 million homes.
• Timeline: Full ramp-up of the project is expected by 2028, but some sites may become operational sooner.
• Locations: Potential data center sites include Wisconsin, Pennsylvania, and Texas, with expansion of existing campuses also planned.
• Oracle's Benefit: This deal is a significant win for Oracle, potentially tripling their current cloud business revenue by $30 billion per year starting in fiscal year 2028. However, the impact on margins and cash flow remains a key question.
• Brody Ford Quote: "Three years ago if you said Oracle [is a] major cloud player, people would make fun of you. Today, that is not the case."

Impact on Chipmakers: NVIDIA and AMD

• Beneficiaries: Bloomberg Intelligence suggests that NVIDIA and AMD will benefit from this infrastructure buildout.
• NVIDIA's Dominance: NVIDIA currently dominates the compute market for AI infrastructure, but AMD is increasingly gaining ground.
• AMD's Gains: AMD is expected to generate $3 to $4 billion in revenue from Oracle's orders of 130,000 MI350X GPUs in the second half of the year.
• Oracle's Strategy: Oracle is unique among the top five cloud providers in not having a custom chip program, meaning they rely on general-purpose GPUs (GPGPUs) from NVIDIA and AMD.
• Kunjan Sobhani Quote: "One gigawatt approximately today's term equates to $60 billion of CapEx which influences everything from chips to hardware."

Market Reaction and the Insatiable Demand for Computing Power

• Market Signal: The deal signals that the demand for computing power is "insatiable."
• Long Runway: The runway for investment in semiconductors and cloud infrastructure, particularly for AI, is still long.
• Energy Implications: The energy sector and infrastructure will need significant investment to support these expansions.
• U.S. Advantage: The U.S. is positioned to lead the AI race due to its relatively cheaper and self-sufficient energy production compared to regions like Europe.
• Janet Hsieh Quote: "The significance of this deal is demand for computing power is insatiable... the runway is still long and the incentive to invest is still high."

Tax Bill and Export Controls

• House Vote: The U.S. House of Representatives is on track to vote on Donald Trump's tax and spending bill.
• Clean Energy Tax Credits: Concessions were made regarding the phaseout of clean energy tax credits, particularly around solar and wind subsidies, to secure Republican votes.
• China Trade: The U.S. is easing export license control requirements for chip software design sales to China.
• EDA Software: Restrictions on EDA software sales to China are being rolled back, allowing companies to support their customers.
• Trade Negotiations: This easing is part of ongoing trade negotiations between the U.S. and China.
• Huawei's Access: Concerns exist that Huawei could use EDA software to close the gap in chip design capabilities.
• Peter Elstrom Quote: "You have export controls which are typically national security issues being traded as part of negotiations."

CoreWeave and NVIDIA's Newest AI System

• First Recipient: CoreWeave will be the first company to receive the latest AI system based on NVIDIA's newest high-end chip.
• Dell's Role: Dell delivered the first GB200 NVL72 servers to CoreWeave.
• Performance Upgrade: The Blackwell architecture offers a significant performance upgrade (approximately 50%) in terms of token processing.
• Backlog Conversion: Having the latest NVIDIA chips first helps CoreWeave with faster backlog conversion.
• Infrastructure Super Cycle: The infrastructure super cycle is playing out, driven by announcements like the Oracle-OpenAI deal and CoreWeave's deployment of new NVIDIA chips.

Jobs Data and the Technology Sector

• June Jobs Report: The U.S. added 137,000 jobs in June, but the report was mixed.
• Government Employment: A significant portion of the job gains (73,000) were in government employment, potentially due to seasonal adjustment issues.
• Tech Sector Losses: The tech sector experienced job losses in computer manufacturing, semiconductor manufacturing, and web search portals/hosting.
• Rate Cut Expectations: The jobs report reduces the urgency for the Federal Reserve to cut interest rates.
• Innovation and Private Equity: Innovation and private equity are key drivers of high-paying jobs in the U.S., but private equity deal activity is currently down.
• Talent Poaching: There is significant talent poaching in the software and AI fields, with executives demanding outsized pay packages.
• Executive Index: Per Spoke Partners uses an executive index to identify and recruit top talent in the software market.

Seed Stage Investing and AI

• Focus: Encore Capital Group focuses on seed investing, writing the first institutional check into companies.
• AI Undercoat: The firm is excited about AI and backs founders with a deep belief in new customer segments or companies with AI as their "undercoat."
• Larger Rounds: Seed rounds are getting larger, with the average in Q1 exceeding $3 million.
• AI Native vs. AI Adjacent: AI is rapidly shifting from being a sector to encompassing every startup.
• Ideal Customer: Seed-stage companies need to focus on a particular customer segment and own their workflow end-to-end.

Meta's Threads and Social Media

• Threads' Growth: Threads has grown to 350 million monthly users since its debut.
• Identity Crisis: The platform is still working on finding its own identity among social media platforms.
• Lack of News Focus: Threads missed an opportunity to plant its flag on the news side.
• Platform for Ideas: Meta positions Threads as a platform for exchanging ideas and stoking interaction.

Social Media and Immigration

• Scrubbing Accounts: Immigration lawyers are advising clients to scrub social media accounts of controversial or politically sensitive topics.
• Potential Repercussions: Posts could be used to block entry to the U.S. or as grounds for removal.
• Influencer Concerns: Influencers are being advised not to post on "hot button" topics for fear of potential repercussions.
• Hassan Piker Example: Twitch streamer Hassan Piker was detained and questioned when entering the U.S.
• Conservative vs. Left-Leaning: Concerns exist that the gap between conservative and left-leaning influencers could widen due to fears of speaking about certain topics.

Conclusion

The broadcast covers a range of topics, from significant investments in AI infrastructure and their impact on chipmakers to the complexities of the job market and the evolving landscape of social media. Key takeaways include the insatiable demand for computing power driving massive infrastructure projects, the strategic importance of energy resources in the AI race, and the increasing scrutiny of social media activity in immigration processes. The discussion also highlights the dynamic nature of the technology sector, with companies like Oracle reinventing themselves as major cloud players and startups navigating the challenges of a rapidly evolving AI landscape.