Citron Raises Amazon Price Target to $300, Amazon May Challenge Nvidia’s AI Chip Market Position

TradingKey - Citron Research founder Andrew Left recently raised Amazon's ( AMZN) price target to $300, which, according to his estimates, represents nearly 30% upside from Thursday's closing price. The rationale this time is not just that "AWS is still growing," but rather a more direct view of Amazon's proprietary chip business as a massive, undervalued asset.

Left believes that Amazon is leveraging its custom chips, AWS cloud business, and expanding AI revenue to realize a profit trajectory previously underestimated by the market, positioning Amazon as one of the most formidable threats to Nvidia ( NVDA) in the AI chip market.

Why Citron Is Suddenly Bullish on Amazon

Andrew Left's core logic for being bullish on Amazon this time is that "the market has not yet fully priced in the AWS chip business."

Amazon CEO Andy Jassy disclosed in his latest shareholder letter that AWS's AI business has reached an annualized revenue run rate of over $15 billion, while its custom chip business has exceeded $20 billion; if the company begins selling its proprietary chips to third parties in the future, annualized revenue for this business line could potentially reach $500 billion.

What Left sees is that this profit stream is rapidly taking shape, but it has not been fully reflected in Wall Street's models.

To Left, Amazon is not just an e-commerce or cloud computing company, but a massive platform that can bundle AI computing power, chips, cloud services, and customer demand together.

Jassy also explicitly noted in the letter that the price-performance ratio of Trainium3 has improved by 30% to 40% over its predecessor, and a significant portion of Trainium4 capacity has already been pre-ordered. More importantly, he directly compared AWS's proprietary chip roadmap to how Amazon entered the CPU market with Graviton and gradually eroded Intel's market share.

Left's bullish case for Amazon is not about whether the company 'can make chips,' but rather whether it can turn chips into a new profit lever for AWS.

Why Amazon has the opportunity to challenge NVIDIA's dominance

Amazon's most realistic entry point for challenging Nvidia is not the most powerful GPUs for training large models, but rather inference and enterprise-level workloads that prioritize price-performance and are better suited for cloud platforms to handle internally.

Jassy was quite direct in his shareholder letter: customers want better price-performance, and while AWS will continue to be the best place to run Nvidia chips, "new changes have already begun." This means Amazon is not looking to push Nvidia out immediately, but is instead shifting a portion of demand toward its own Trainium and Graviton ecosystems.

Meanwhile, Jassy emphasized that the value of Trainium and Inferentia lies in their lower costs and superior price-performance, helping customers perform complex computations and process massive datasets more affordably.

Jassy also stated that AWS is already, via cloud services, to OpenAI, Anthropic, and Apple providing chip capabilities, Uber is also using Amazon's custom chips to accelerate computing and AI model training. This indicates that Amazon's chips are more than just internal trials and are starting to build usage habits in real-world customer scenarios.

This "cloud-first, then gradual external rollout" model is where Amazon has a real opportunity to erode Nvidia's market share.

Why is he bearish on Nvidia?

Left's bearish stance on NVIDIA does not imply that NVIDIA will lose its dominance immediately; rather, he believes that NVIDIA's most comfortable high-growth phase may have begun to transition into a more pronounced competitive cycle.

According to Reuters, NVIDIA is facing pressure not only from OpenAI, Meta, and other such customers' in-house chip initiatives; an increasing number of major clients are attempting to shift a portion of their AI inference workloads to their own ASICs. One analyst even predicts that as in-house ASIC projects scale up by 2027, NVIDIA could begin to see its market share erode.

A more realistic point is that NVIDIA will not collapse immediately just because competition has emerged; it still maintains significant advantages in orders and its ecosystem. Reuters previously reported that NVIDIA will sell 1 million GPUs to AWS by 2027, showing that Amazon remains reliant on NVIDIA in the short term.

Consequently, Left's perspective is more akin to suggesting that "NVIDIA's optimal pricing power is declining" rather than "NVIDIA will be rapidly phased out."

Within the AI chip landscape, AWS, Google, Meta, and OpenAI—these major customers are all developing their own chip pathways. NVIDIA must defend not only its market share but also its leverage in customer renegotiations over pricing and compute efficiency.

Therefore, the logic behind Left's bearish stance on NVIDIA is not centered on the "chips being inadequate," but on the fact that "as customers start making their own chips, NVIDIA's bargaining power will be gradually squeezed." This is also why he describes Amazon as the "most serious threat to NVIDIA."

Amazon is not a pure-play chip startup; it is a giant with its own cloud, customer base, and data center scale. Once this in-house chip roadmap matures, the pressure on NVIDIA will be greater than that from typical competitors. This assessment is an inference based on the chip revenue disclosed by Jassy, customer usage patterns, and market trends regarding ASICs.

What does this mean for investors?

For Amazon, Citron raising its price target to $300 is more of a confirmation that the market had previously underestimated the potential of the AWS chip business.

Today's Amazon is more than just e-commerce and cloud; it is taking another step toward becoming an "AI infrastructure company." Jassy has placed AI revenue, chip revenue, and future third-party sales at the forefront, effectively outlining a second growth curve for investors.

Left's bullish stance on Amazon is essentially a bet that this chip business will drive Amazon's valuation structure further upward in the future.

For Nvidia, the impact of this news is not the "immediate loss of dominance," but rather an indication that more hyperscale cloud providers are internalizing their AI compute capabilities.

Broadcom The long-standing chip partnership with Google already demonstrates that custom silicon is not a move unique to Amazon, but a common direction for the entire cloud computing industry. The faster Amazon moves along this path, the more Nvidia faces not just a single competitor, but a cohort of deep-pocketed giants with their own customers and data centers.

For investors, the key focus is not whether Nvidia's market position will be replaced, but whether Amazon AWS's chip business can truly scale into a large, independently priced business as Jassy envisions.