AI bubble: Data centre cancellations are sky rocketing | Ed Zitron

The Viability of OpenAI and Oracle's Data Centers

Current State of Data Centers

• The only tenant for the data centers is OpenAI, raising concerns about its sustainability over the next two and a half years.

• There are significant financial issues with banks losing hundreds of millions on unbuilt data centers, indicating a potential bubble in the market.

Investor Sentiment and Financial Outlook

• Recent changes in Oracle's credit outlook to negative by S&P Global suggest investors are reassessing their enthusiasm for tech investments.

• Concerns arise regarding Oracle's ability to afford its agreements, as it faces declining software business performance and revenue reporting issues.

Challenges in Funding and Construction

• Blue Owl Capital’s withdrawal from a $10 billion deal highlights difficulties in securing funding for Oracle’s projects.

• Banks are questioning the profitability of long-term data center projects, leading to tighter lending conditions.

Economic Viability of AI Infrastructure

• The economics surrounding AI infrastructure remain unclear; there is uncertainty about GPU costs and profitability.

• Analysts note that investment interest has shifted primarily towards Asian banks due to rising skepticism among US banks regarding project viability.

Delays and Project Feasibility

• Major projects like Stargate Abene face significant delays, with construction timelines being pushed back repeatedly.

• Investors express concern over Oracle's ability to deliver on commitments, especially given OpenAI's reported losses which could impact payment capabilities.

Investment Dynamics in OpenAI

SoftBank's Financial Position and Investment Strategy

• Despite financial constraints, SoftBank is reportedly considering a $30 billion investment in OpenAI. The discussion raises questions about their capacity to fund such an amount given their recent financial maneuvers.

• SoftBank's current financial situation is precarious; they had to sell Nvidia stock and take out a $15 billion margin loan on ARM stock just to meet previous obligations to OpenAI.

• The potential for a $30 billion investment from SoftBank seems questionable as it may require further asset liquidation, including Alibaba stock, indicating that they are nearing the limits of their financial resources.

Venture Capital Landscape and Valuation Concerns

• The venture capital landscape appears strained; traditional VCs can only afford smaller investments (around $5 billion), making it difficult for them to keep up with the funding demands of companies like OpenAI.

• There are rumors that OpenAI plans to go public by the end of 2026, which would mean these last rounds of funding could be among the final opportunities for private investment at inflated valuations.

Market Reactions and Circular Financing

• Current market dynamics suggest that major players like Amazon, Microsoft, and Nvidia are stepping in because traditional venture capital cannot meet the high funding requirements anymore.

• Reports indicate fluctuating figures regarding potential investments from various tech giants, raising skepticism about whether these funds are real or merely speculative promises.

Transparency Issues with OpenAI

• There are concerns regarding transparency at OpenAI; sources suggest that investors lack clear information about its financial performance since reaching a valuation near $100 billion years ago.

• Speculation exists that OpenAI might be withholding critical operational data from potential investors while requesting substantial funding amounts, complicating trust in their business model.

Implications of Investment Strategies

• Companies like Amazon have significant contracts with OpenAI but face criticism for engaging in what some describe as "circular financing," where money flows back into the same entities without creating genuine value.

• The conversation touches on ethical considerations surrounding investments aimed solely at sustaining customer relationships rather than fostering real growth or innovation within those companies.

• Historical context is provided regarding Nvidia's past claims of investing heavily in OpenAI which did not materialize, leading to skepticism about future commitments from them or other firms.

Concerns Over AI Investment and Data Center Viability

Investment Skepticism in AI Companies

• The speaker reflects on the unfulfilled promise of a $10 billion investment in Anthropic, questioning the short-term memory of stakeholders regarding past announcements.

• There is a pattern of inflated fundraising claims from companies like OpenAI and Anthropic, suggesting they are leaking information to create an illusion of financial strength.

• The speaker expresses confusion over the purpose and sustainability of these investments, highlighting concerns about profitability and future viability.

Data Center Development Challenges

• A trend is noted where announced data centers are not being built as expected; 41 data centers have been canceled or delayed recently due to various issues.

• Reasons for delays include local opposition and power grid limitations, but "unfavorable economics" stands out as a significant concern among investors.

Economic Viability of Data Centers

• The term "unfavorable economics" raises questions about its true meaning in the context of data center financing and operations.

• Many GPU data centers are likely unprofitable; worsening debt conditions could make them financially unsustainable even if they were initially viable.

Debt Implications for Construction Projects

• Oracle's $38 billion loan for data centers involves interest-only payments until construction completion, raising concerns about long-term financial health.

• The reliance on OpenAI as the sole tenant poses risks; doubts are cast on OpenAI's longevity given current funding challenges.

Risks Associated with Rapid Technological Changes

• The economic model for these projects appears fragile; any increase in loan costs could render repayment impossible due to insufficient revenue streams.

• Building new facilities is complicated by rapid technological advancements that may render existing infrastructure obsolete before it becomes profitable.

Financial Strain from Poor Margins

• Reports indicate that Oracle's GPUs have negative gross margins, complicating their ability to repay loans amidst high operational costs associated with cooling and maintenance.

• Overall economic conditions surrounding these projects are precarious; slight changes can lead to severe financial repercussions.

Construction Challenges in Data Centers

Overview of Construction Issues

• The speaker discusses the prevalence of budget overruns in construction projects, noting that they have never seen a project come in under budget.

• Specific challenges are highlighted for data centers being built in Texas, including severe weather and local community issues that drive up costs.

Rising Costs and Financial Implications

• There is a significant shortage of materials like steel for power transformers, contributing to increased costs across the board due to high demand from data center construction.

• Concerns are raised about how companies like Blue Owl and JP Morgan will manage rising construction costs and delays, with no clear answers provided.

Business Journalists Learning Curve

• The speaker notes that business journalists are now having to learn about finance and construction as these topics become increasingly relevant to reporting on tech companies.

• A prediction is made regarding OpenAI's financial sustainability over the next two and a half years, suggesting they may not survive without significant changes.

Financial Transparency Concerns

• The speaker expresses skepticism about OpenAI's reported financial figures, suggesting they may not be fully transparent about their costs.

• Comparisons are drawn between OpenAI’s financial trajectory and other companies like Anthropic, indicating similar patterns of unsustainable losses.

Market Realities and Predictions

• The discussion highlights concerns over the viability of major tech firms given their massive debts and operational losses.

• Oracle's financial situation is scrutinized; it faces substantial debt obligations while relying heavily on revenue from clients like OpenAI who are also struggling financially.

Future Outlook for Data Centers

• A report predicts that up to 50% of announced mega data centers may never be built, raising questions about the future of AI infrastructure.

• The implications for both the AI industry and broader economic conditions are discussed, emphasizing uncertainty around which projects will proceed.

The Viability of AI Data Centers: Challenges and Concerns

Financial Viability and Industry Optimism

• The speaker expresses skepticism about the optimism surrounding a 50% success rate for AI data center projects, suggesting that the required funding exceeds what has been raised (e.g., $178.5 billion by 2025 in the US).

• There is concern that if many proposed data centers fail to materialize, it will negatively impact the AI industry's credibility, necessitating stronger proof of revenue potential and efficacy.

• Companies like Salesforce are struggling to demonstrate substantial revenue growth, indicating broader issues within the industry regarding tangible success.

Perception of AI's Reality

• The speaker argues that unbuilt data centers and failed debt deals serve as concrete evidence undermining confidence in the AI sector.

• This situation could lead to stock sell-offs as investors react to negative reports about data center viability.

Investment Risks and Market Sentiment

• Banks may hesitate to finance new data center projects due to emerging risks, with Blackstone being noted as an outlier willing to invest despite these concerns.

Infrastructure Challenges

• The lengthy lead time (up to eight years) for connecting new data centers to power grids complicates financial planning and project feasibility.

• A stark contrast exists between the time needed for infrastructure development versus debt repayment timelines, raising questions about profitability.

Technical Hurdles in Power Supply

• Issues arise not just from generating power but also from delivering it effectively; logistical challenges can delay operations significantly.

• Local opposition against projects like Stargate Abilene highlights community concerns over disruption caused by construction efforts.

Safety Regulations and Delays

• Strict safety regulations must be adhered to during power testing phases; failures can result in catastrophic outcomes, further delaying project timelines.

Economic Feasibility of Future Projects

• Even if funding were available, significant doubts remain regarding whether necessary infrastructure will be completed on time or at all.

• Oracle's ambitious plans for building capacity raise questions about economic viability given projected costs exceeding $150 billion without clear returns on investment.

This structured summary captures key insights from the transcript while providing timestamps for easy reference.

Data Centers and Local Employment: A Critical Analysis

The Impact of Data Centers on Local Economies

• Data centers are primarily staffed by highly paid, specialized talent that is not sourced locally; instead, workers are flown in at a premium cost.

• The construction of data centers in Washington state does not contribute to local job creation as the workforce is external.

• These facilities often receive tax abatements, raising questions about their economic benefits to the local community.

• In California, similar trends are observed where data centers do not provide significant employment opportunities for residents.