NVIDIA Proves AI Profitability… The Worst is Over

NVIDIA's recently announced earnings showed both revenue and earnings per share (EPS) surpassing market expectations. However, they did not reach the level of 'earnings surprise' that NVIDIA has recorded in the past, marking a solid performance.

This figure aligns with the upper end of market expectations, which ranged from $103 billion to $105 billion, thereby meeting market anticipation.

Through this earnings announcement, NVIDIA demonstrated that AI is generating tangible profits, evidenced by customer distribution and growth. Beyond traditional big tech companies, NVIDIA has expanded its customer base to include frontier labs, national sovereign AI projects, and various cloud companies, proving the commercial viability of AI technology.

This figure significantly exceeds initial market forecasts of mid-40%, suggesting that additional growth could be possible if potential bottlenecks are resolved.

This is a steep growth, considering it was 32% six months ago and 42% last quarter. This is an important indicator showing that even small and medium-sized enterprises with financial constraints are increasing their AI adoption.

Memory semiconductor purchase commitments increased significantly from $250 billion to $150 billion compared to the previous quarter. This indicates that customers are actively moving to secure memory semiconductor supply, demonstrating the surging demand for related components alongside the growth of the AI market.

The projected increase in demand, once expected to be 10-20 times, is now mentioned to potentially surge up to 100 times.

Beyond the earnings figures themselves, the underlying growth potential and spread of the AI industry hold more significant meaning.

This serves as crucial evidence to alleviate past concerns about CAPEX sustainability. The AI market itself is growing, leading to a positive long-term outlook for NVIDIA.

This suggests limitations in disclosing detailed debt information beyond financial statements, making a clear answer difficult.

Regarding Rubin Ultra, there was no specific mention of memory specifications such as HBM (High Bandwidth Memory) capacity. The current main issue is production bottlenecks occurring during the introduction of Rubin, following Blackwell. The lack of smooth production has been identified as a core problem.

This suggests an acceptance of rising semiconductor purchase prices.

This appears to be a result of a product mix change strategy driven by price increases.

As a result, it helped alleviate many market worries.

The US economy is in an ambiguous state, showing slowing employment and weaker-than-expected consumption patterns. In this situation, raising interest rates could cause economic problems, creating a neutral scenario where rates can neither be raised nor lowered. A complex interplay between interest rate policy and economic vulnerabilities appears to be at play.

US Treasury yields serve as a crucial threshold for the survival of AI supply chain companies. Unless the 10-year Treasury yield reaches 4.9% to 5% or the 30-year yield approaches 6%, most AI supply chain companies (about 70-80%) are expected to be fine. However, companies with excessive debt, insufficient backlogs, or high dependence on specific customers could become 'weak links.' These risk factors should be analyzed through cases like Oracle and OpenAI.

Despite debt issues, neo-cloud companies are maintaining sound financial health based on robust performance. For instance, Queryb, despite significant debt, has secured a backlog more than three times its size, and Navius reduces financial risk by receiving 40-50% customer advance payments. NVIDIA's strong support plays a crucial role in the financial stability of these companies.

Risks from 'weak links' like Oracle or OpenAI are expected to have a limited impact on the overall AI infrastructure market. The robust growth of the remaining 75% of the market, including sovereign AI and enterprise, is more significant than these links, which account for about 25%. NVIDIA's earnings report today confirms the growth of these mainstream markets, suggesting that the current stock price has already priced in these risks, and there's no need for an exaggerated reaction. The overall AI infrastructure market possesses strong growth momentum.