Intel and Broadcom CEOs warn AI sector faces capital access hurdles
Executives at two major chipmakers highlight funding constraints and supply chain imbalances as key risks for the artificial intelligence industry.
Key points
- Intel and Broadcom leaders cite capital access as a growing challenge for AI firms.
- Chip supply shortages may ease by 2028, but capacity gaps could persist until 2030.
- Many AI labs rely on pilot projects rather than sustainable enterprise revenue streams.
Intel Corporation chief executive Lip-Bu Tan and Broadcom Inc. executive Hock Tan identified capital access as a significant potential challenge for the artificial intelligence sector. The two leaders shared these observations during a financial conference held in Singapore today, highlighting economic pressures facing the industry.
Both executives represent two of the most critical chip suppliers in the AI ecosystem. Intel’s central processing units remain in high demand because AI agents require them to execute tools. Broadcom supplies network processors and custom AI accelerators that are essential for data center infrastructure and model inference.
Supply Chain and Market Outlook
Demand for specific data center chips currently exceeds available supply. Intel disclosed last month that it possesses the capacity to meet only 50 percent of incoming orders. This shortfall underscores the intense competition for hardware resources among technology companies building and deploying artificial intelligence systems.
Hock Tan predicted that the current chip shortage would ease by 2028 or 2029. However, he cautioned that the imbalance between the demand for computing capacity and the industry’s ability to supply it could persist into 2030. This suggests a prolonged period of hardware scarcity for AI developers.
Financial Pressures on AI Labs
Tan addressed the steep costs associated with training large language models, focusing specifically on OpenAI Group PBC. Broadcom assisted the AI developer in designing its first custom inference chip earlier this year. Tan noted that OpenAI remains unprofitable because of heavy spending on training activities.
He stated that the company’s current financial picture does not indicate anything except the potential for significant revenue growth from frontier generative AI models. This comment highlights the gap between early-stage investment and long-term profitability in the sector.
Revenue Models and Investor Expectations
Tan observed that a sizable portion of AI labs’ enterprise revenue currently originates from pilot projects. He compared this dynamic to a child playing in a video arcade with limited tokens, noting that users often finish their budget quickly without learning substantial outcomes.
The executive added that AI labs should acquire more customers who can realize a tangible return on their inference spending. He also commented on sharp fluctuations in AI stocks, remarking that many investors set extreme expectations only to find out that it is almost like they are in a casino.
What should you do?
Security and technology professionals should monitor supply chain communications from chip vendors for updates on capacity constraints. Organizations relying on AI services should evaluate the financial stability of their providers, particularly those dependent on pilot revenue. Teams should also review their own infrastructure plans to account for potential hardware shortages extending into the next decade.
Background: Large language models
Large language models are a specific type of neural network trained on text. Generative AI is the broader umbrella term for any system that creates new content, including images and code. Not all generative AI uses large language models, but most modern text generators do.
What to do and how to stay safe: Intel
- Review current contracts with AI service providers to understand their revenue models and financial health.
- Diversify hardware suppliers to mitigate risks associated with single-source chip shortages.
- Monitor vendor announcements for updates on capacity constraints and delivery timelines.
- Evaluate internal AI projects for cost efficiency and realistic return on investment metrics.
Step-by-step guide: Prompt Injection Warning Signs: How to Spot AI Manipulation
General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
When do experts expect the AI chip shortage to ease?
Broadcom executive Hock Tan predicted the shortage would ease by 2028 or 2029, though capacity imbalances may persist into 2030.
What is Intel's current capacity for meeting orders?
Intel disclosed last month that it has the capacity to meet only 50 percent of orders for its central processing units.
Why are AI labs struggling with profitability according to Hock Tan?
Tan noted that companies like OpenAI are losing money due to high training costs and reliance on pilot projects rather than sustainable enterprise revenue.
