SI Agent News
DevelopingOpenAIChips

OpenAI president reportedly asked Musk and Brin for AI compute

A report describes OpenAI's Greg Brockman seeking hardware resources from rivals, highlighting a severe shortage of memory chips and processing power.

87 percentMicron's gross margin this past quarter

The short version

Why it matters

This interaction highlights that AI compute is so scarce that even major rivals are being asked to share resources. It also signals that memory, not just processing power, is becoming the primary physical constraint on AI development.

A request for resources

According to a feature by Colossus, OpenAI president Greg Brockman reportedly approached both Elon Musk and Sergey Brin to ask for compute. The exchange occurred during a dinner involving several of the world's most influential technology executives. These leaders were gathered to receive national medals from President Trump.

Musk described the interaction as Brockman hounding him for resources during the meal. Brin echoed this sentiment, confirming that Brockman had asked him for compute as well. The conversation took place in an auditorium where the executives were held by the Secret Service.

The report suggests that the scarcity of hardware is reaching a point where established rivals are seeking favors from one another. While the report does not state whether Musk or Brin agreed to the request, it highlights the intense pressure on OpenAI to secure enough power to continue scaling.

The memory bottleneck

While the conversation focused on compute, Musk noted that the lack of memory chips was an even larger issue. He characterized the memory market as the worst supply chain price shock he has ever witnessed. This indicates that the physical components required to store and move data are becoming as significant a hurdle as the processors themselves.

The data supports this observation of a significant price shift in the hardware market. In mainstream PCs, the cost of memory and storage rose from 16% to 42% of the total parts cost. This represents a 2.6 times increase in the share of the total cost, adding about $224 to the price of a $699-tier PC.

This shift suggests that the demand for high-bandwidth memory is outstripping the available supply. As AI models grow larger, the need for efficient memory becomes a primary rate-limiting factor in the scaling equation. The hardware market is currently struggling to keep pace with the rapid expansion of AI infrastructure.

Market dynamics and margins

The financial health of memory manufacturers reflects the extreme demand currently facing the industry. Musk expressed disbelief at Micron’s 87 percent gross margin reported this past quarter. Such high margins are a direct result of the supply-demand imbalance in the memory sector.

However, the ability of these companies to expand production remains a point of contention among industry leaders. Satya Nadella commented on Micron’s capital expenditure plans, stating, "They’re not really putting the money back." This suggests a potential hesitation in reinvesting profits into new capacity at the current pace.

For device makers, these costs are becoming increasingly difficult to absorb. The jump in memory costs for mainstream PCs shows that the AI-driven demand is bleeding into the broader consumer electronics market. Manufacturers must navigate these rising costs while trying to maintain viable price points for end users.

The cost of scaling

The interaction between Brockman, Musk, and Brin illustrates the high stakes of the current AI race. Because compute is so scarce, even a company that has been involved in legal disputes with a rival might be forced to consider cooperation. Musk even remarked on the audacity of the request, noting that Brockman had previously been involved in a lawsuit against OpenAI's CEO.

The scarcity of Blackwell chips also plays a role in this landscape. Musk noted on X that there will be a total of around 1.2M Blackwells at Colossus 2 by the end of the year. This limited supply creates a environment where every available unit of compute becomes a strategic asset for the major labs.

Ultimately, the conversation highlights a transition from a software-driven competition to a physical resource competition. The winners of the next phase of AI development may be determined by who can secure the most stable supply of memory and processing power. The

What to watch

Try it

Look up the current price of a mid-range laptop and check if the memory specifications seem disproportionately expensive compared to previous years.

Sources and links

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

Listen to the AI Wire headlines

Listen to the AI Wire

Listen to our daily AI shows every morning: on your commute, in the shower, wherever you are

Listen to the daily shows

First reported by FourWeekMBA · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

More AI news