Key points
- Intel can currently meet only half of its chip orders.
- Broadcom expects supply shortages to ease by 2028 or 2029.
- Many AI labs rely heavily on pilot projects rather than full deployments.
Chipmakers are signaling that the artificial intelligence industry faces significant hurdles in accessing capital and securing hardware. Intel Corp. chief executive Lip-Bu Tan and Broadcom Inc. CEO Hock Tan highlighted these issues during a financial conference in Singapore today. Both companies are critical suppliers of processors and accelerators needed to run AI workloads.
By the numbers
- 50%: The portion of Intel's current orders that it has the capacity to fulfill.
- 2028-2029: The timeframe when Broadcom expects chip supply shortages to begin easing.
- 2030: The year until which an imbalance in computing capacity could persist.
What the figures mean
Intel’s central processing units are in high demand because AI agents require them to execute tools. According to SiliconANGLE, Intel recently disclosed that it can only satisfy half of its orders. Broadcom supplies network processors and custom AI accelerators. Hock Tan predicted that while the immediate shortage would ease by 2028 or 2029, the gap between demand for computing power and supply might last until 2030.
Tan also addressed the high costs of training large language models. He noted that OpenAI Group PBC is currently losing money due to heavy training expenditures. Broadcom helped design OpenAI’s first custom inference chip earlier this year. Tan stated that the losses indicate potential revenue growth rather than failure.
The Financial Times reported last month that Anthropic PBC achieves gross margins exceeding 80% when excluding training costs and partner revenue shares. Anthropic is reportedly on track for positive adjusted operating income for a second consecutive quarter. Tan observed that much of AI labs' enterprise revenue comes from pilot projects. He compared this to children spending tokens quickly at an arcade without learning anything lasting.
Tan advised AI labs to secure more customers who can realize a return on inference spending. He also commented on volatile AI stocks, saying investors often set extreme expectations akin to gambling in a casino. That suggests the market may need to adjust its timeline for profitability and hardware availability.
What should you do?
Managed service providers should review their hardware procurement strategies given the predicted multi-year supply constraints. Organizations should evaluate whether their AI initiatives rely too heavily on pilot programs rather than scalable deployments. Monitoring vendor capacity reports will help in planning long-term infrastructure investments.
Background: Large language models
The difference between large language models and small language models lies in their parameter counts, training data volumes, and memory footprints. Large models handle complex, open-ended reasoning across many domains, while small models execute narrow tasks efficiently on local hardware with predictable latency and lower resource overhead.
What to do and how to stay safe: Intel
- Audit current chip inventory and lead times to anticipate supply gaps.
- Review AI project budgets to ensure they account for sustained training costs.
- Prioritize clients with clear paths to return on inference spending.
- Monitor vendor announcements for updates on production capacity and timelines.
Step-by-step guide: Shadow AI Explained: 8 Questions Answered for Secure Operations
General security guidance from the BitShield News newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
When will the AI chip shortage end?
Broadcom CEO Hock Tan predicts the shortage will ease by 2028 or 2029, but capacity imbalances may persist until 2030.
How much of Intel's orders can it currently fill?
Intel disclosed last month that it has capacity to meet only 50% of its orders.
Why are AI labs struggling with profitability?
Executives note that high training costs and reliance on short-term pilot projects hinder immediate financial returns.
