ProductNot yet confirmed elsewhere1 publisher2 min readPublished
Broadcom's Hock Tan sees the shortage of some data center chips easing by 2028 or 2029
Broadcom CEO Hock Tan predicted the shortage of some data center chips will ease by 2028 or 2029 and could linger into 2030. Intel's Lip-Bu Tan spoke at the same Singapore event; the report does not include a date for Intel's supply to improve.
The Product Desk
What happened
- Intel disclosed last month that it has capacity to meet only 50% of orders, SiliconANGLE reported.
- Demand for Intel's central processing units is high because AI agents need them to run their tools.
- Hock Tan said pilot projects currently account for a large share of AI labs' enterprise revenue.
Why it matters
- constraint Intel can fill only 50% of its orders, so any CPU plan that leans on a single Intel order needs a fallback for the unfilled half.
- contradiction The executive forecasting a shortage also says money might limit the sector more than chips do, so a supply date alone is an incomplete planning input.
- exposure Labs whose enterprise revenue leans on pilots are exposed if those customers stop when the allotment does, and Hock Tan wants more customers who can show a return.
- decision A team facing a capacity wait can test the open-source models Lip-Bu Tan called "good enough" before it commits to scarce hardware.
Hock Tan's version of a pilot customer is a child in a video arcade. "When you are a kid you go to a video arcade, you play," he said. "Your parents give you $10 of tokens, you finish it in 10 minutes and you haven't learned anything." [8] By his account, users spend the allotment and stop. The lab books that spend as demand.
That matters because Tan also said demand for some data center chips currently exceeds supply [3]. He put the easing at 2028 or 2029 [4] and said the imbalance could persist into 2030 [5].
Capital is the constraint both executives named. Lip-Bu Tan said: "if there's any challenge in the front, it's capital and the cost of capital, and then the other part is in the longer run, people want to see the ROI." [7] Hock Tan's remark about capital concerned the sector as a whole [6]. He also said AI labs should sign up more customers whose inference spending pays them back [10].
The two companies sell different parts. Broadcom supplies network processors and custom AI accelerators [12], and the 2028 or 2029 estimate came from its chief executive. Intel's 50% figure covers its own orders [13].
For a team that has to commit budget on Monday, we'd sort each workload on two questions: whether it depends on the chips in short supply, and whether it shows a return on inference spend today or only in a pilot.
- Scarce chips, proven return: secure capacity first, with a fallback for the unfilled half of any Intel order [13]. - Scarce chips, pilot only: hold the commitment and try an open-source model first. Lip-Bu Tan said Chinese open-source models are proving "good enough" for many budget-conscious users [11]. - No scarce chips, proven return: leave it alone. - No scarce chips, pilot only: keep it small until it shows a return.
The first is the only quadrant where Hock Tan's 2028 or 2029 estimate belongs in the plan.
What to watch
- Whether Intel gives its own supply date or a fill-rate update after last month's disclosure that it can meet 50% of orders.
- Whether Broadcom says which chips its 2028 or 2029 estimate covers.
- Whether AI labs report a shrinking share of enterprise revenue from pilot projects, which Hock Tan said they should.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence40
- Adoption
- Insufficient
- Hype gap
- Insufficient
- Incentives55
- Confidence60
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
Last month, Intel disclosed that it only has capacity to meet 50% of orders.
- [2]
Intel's central processing units are in high demand because AI agents need them to run their tools.
- [3]
Demand for some data center chips is currently exceeding supply.
- [4]
Hock Tan, Broadcom's CEO, predicted at the financial event that the shortage would ease by 2028 or 2029.
- [5]
Hock Tan believes the imbalance between demand for computing capacity and the industry's ability to meet it could persist into 2030.
- [6]
Hock Tan flagged that access to capital "might be the biggest constraint" facing the AI sector.
- [7]
"if there's any challenge in the front, it's capital and the cost of capital, and then the other part is in the longer run, people want to see the ROI."
- [8]
"When you are a kid you go to a video arcade, you play," he said. "Your parents give you $10 of tokens, you finish it in 10 minutes and you haven't learned anything."
- [9]
Hock Tan noted that a sizable portion of AI labs' enterprise revenue currently comes from pilot projects.
- [10]
Hock Tan added that AI labs should acquire more customers who can realize a return on their inference spending.
- [11]
Lip-Bu Tan noted that open-source language models from Chinese labs are proving "good enough" for many budget-conscious users.
- [12]
Broadcom is a major supplier of network processors and custom AI accelerators.
- [13]
By Intel's disclosed figure, the other 50% of orders is beyond its current capacity.
Sources
1 independent publisher whose own reporting we read for this story.
- siliconangle.comIntel, Broadcom CEOs flag AI industry challenges, trends at financial event
1 article · October 12, 2026
Topics and entities
Follow any of these and your For You feed starts watching them — no settings page required.