To The Moon

$Bernstein U.S. Research Fund(BERN)$

Bernstein emphasizes that the AI industry remains in its growth phase, and even as OpenAI advances its in-house chip development plans, GPU technology continues to hold an indispensable position. This has allowed GPUs to regain an advantage in the competitive landscape against ASICs (Application-Specific Integrated Circuits). In practice, the two may form a complementary landscape: $NVIDIA(NVDA)$ graphics cards primarily handle model training, while $Broadcom(AVGO)$ ASICs focus on inference tasks. This further underscores the market's intense demand for computing resources, with all parties actively competing for access.

DA DAVIDSON

DA DAVIDSON describes NVIDIA as the "ultimate investor," a concept drawing on the metaphor of "lenders of last resort." NVIDIA previously played a similar role in Coreweave's IPO, and this investment in OpenAI further exemplifies this trend. From a broader perspective, NVIDIA is viewed as the "central bank" of AI, monopolizing the issuance of core computing power. Here, GPU computing power is analogized to "currency," with NVIDIA defining the "Compute Anchor"—a metric where all AI service costs can be converted into token generation costs per second based on NVIDIA GPUs. This equates to the "minting cost" within the AI ecosystem. NVIDIA also wields a suite of "monetary policy" tools, operating with striking parallels to the Federal Reserve: launching new chips functions like "interest rate cuts" or "quantitative easing," while its GTC conferences resemble "Fed policy meetings." By managing market expectations, NVIDIA maintains system stability—simultaneously driving technological innovation while preventing market overheating.

$Barclays PLC(BCS)$

We are broadly optimistic about NVIDIA's investment in OpenAI. The $100 billion investment is expected to yield a 3-5x return over the long term, generating $300-500 billion in revenue while solidifying NVIDIA's position as OpenAI's preferred supplier. NVIDIA holds a dominant position in AI infrastructure, with a 2026 fiscal year P/E ratio of approximately 30x and a reasonable PEG ratio (close to 2x), aligning its valuation with growth potential. The OpenAI investment represents an "equity stake + commercial partnership" (similar to the CoreWeave case), positioning OpenAI as just one of NVIDIA's key major clients, with manageable risk. With free cash flow margins of 40%-50%, NVIDIA possesses ample future cash flow, making the OpenAI investment an optimal choice to expand its market and accelerate new product launches. General-purpose chips will continue to handle the majority of OpenAI's workloads. The collaboration to deploy a 10GW system is valued at over $350 billion (3.5 times the scale of Broadcom's comparable project), representing incremental business whose value to NVIDIA is currently underestimated by the market.

$Bank of America(BAC)$

From a demand perspective, both custom and general-purpose chips are experiencing dual peaks, though general-purpose chips will remain dominant in the long term. The industry's focus on whether custom ASICs represent the future of AI computing intensified following Broadcom's $10 billion partnership with OpenAI. This was further fueled by slightly weaker computing business data in Nvidia's earnings report. The latest announcements indicate that AI computing demand is exploding across the board, with both custom ASICs (specialized chips) and general-purpose chips reaching peak demand. Long-term trends suggest that general-purpose chips will continue to handle the vast majority of OpenAI's (unlisted, with broad business coverage) workloads at least until the end of this decade. Custom ASICs will not completely dominate AI computing.

NVIDIA's incremental business: massive scale, pure growth, representing over $350 billion in value. According to NVIDIA's Q2 FY2026 earnings report, each 1GW (gigawatt) of system installations can generate over $35 billion in revenue for NVIDIA. The collaboration with OpenAI to deploy 10GW of NVIDIA systems represents over $350 billion in value. Compared to Broadcom's upcoming ASIC project for OpenAI in the second half of next year (valued at $10 billion), NVIDIA's business scale is 3.5 times larger ($35 billion vs. $10 billion).

$Morgan Stanley(MS)$

Based on NVIDIA's disclosed "cost structure per gigawatt (GW) of AI computing power" (approximately $50-60 billion per GW, with $35-40 billion flowing to NVIDIA), a 10-GW scale is projected to generate $350–400 billion in long-term revenue for NVIDIA, highlighting the financial potential of this collaboration.

However, it emphasizes the "uncertainty" surrounding the scale. When compared to OpenAI and Oracle's "over $300 billion collaboration," the revenue scale corresponding to Nvidia's investment is deemed "not a certainty," but rather "part of a long-term bullish logic" (rather than an inevitable outcome). The largest currently operational GPU cluster only reaches the "hundreds of megawatts (MW)" level (1GW = 1000MW), showing a significant gap from the target of "10GW." This indirectly reflects the difficulty or extended timeframe required to achieve such scale, further undermining the "certainty of revenue."