Hyperscalers are aggressively expanding custom silicon capabilities to optimize AI inference costs, even as Nvidia Corp (NASDAQ: NVDA) maintains its dominant position in the AI chip market. This strategic shift signals a move toward greater infrastructure control by major technology firms.

According to BNP Paribas Equity Research senior analyst Karl Ackerman, presentations from Meta Platforms Inc (NASDAQ: META) and Alphabet Inc’s Google at the Hot Chips conference underscore a growing industry focus on efficiency. The firms are building proprietary hardware not to replace Nvidia overnight, but to lower the cost of inference as demand scales.

Custom Silicon Roadmaps

The announcements highlight how quickly the largest cloud companies are broadening their AI hardware strategies:

• Meta: Detailed the next phase of its Meta Training and Inference Accelerator (MTIA) roadmap. The upcoming MTIA 400 chip supports recommendation systems and generative AI models. It delivers 3 PFLOPS of FP16 compute—roughly five times the performance of its predecessor—and expands high-bandwidth memory capacity by 33% to 288 GB.
• Google: Focused on networking infrastructure for its latest TPU platform. Its Virgo Network can connect 134,000 TPU 8t chips with up to 47 petabits per second of non-blocking bandwidth. This enables more than 1.6 exaFLOPS of compute while maintaining near-linear scaling, with architecture designed to eventually support up to one million TPU chips.
• Amazon: Continues investing in proprietary AI processors through its Trainium and Inferentia families, reflecting a broader industry push toward custom silicon alongside merchant GPUs.

What the Numbers Show

The data reveals a divergence between training and inference strategies. While Nvidia dominates cutting-edge AI training, hyperscalers are optimizing for inference economics. Meta’s MTIA 400 delivers five times the performance of its predecessor with only a 33% increase in memory capacity, indicating a focus on compute density rather than raw memory expansion. Similarly, Google’s ability to scale to 1.6 exaFLOPS across 134,000 chips highlights the importance of networking bandwidth in achieving linear scaling for large models.

Investment Implications

For investors, this distinction matters. Nvidia remains the primary supplier for training workloads, but the economics of running AI services are encouraging tech giants to own more of the underlying hardware stack. Custom silicon should be viewed less as an immediate threat to Nvidia and more as a measure of how hyperscalers plan to balance dependence on the AI leader with greater control over their own infrastructure costs.

Wall Street analysts project Nvidia Corp. (NASDAQ: NVDA) could add nearly $7 trillion to its market capitalization, reaching a valuation of approximately $12.5 trillion. This potential leap would create market value equivalent to roughly 14 Intel Corp.-sized companies.

Following the chipmaker’s recent earnings report, brokerages raised their price targets. Raymond James set the highest major brokerage target at $515 per share, up from Nvidia’s current trading price of around $226.

The Math Behind Nvidia’s Next Trillion-Dollar Leap

The projected valuation shift highlights the scale of analyst optimism surrounding the semiconductor giant. If Nvidia reaches the $515 target, its market capitalization would climb from its current level of roughly $5.47 trillion to about $12.5 trillion.

To contextualize this growth, Intel Corp. (NASDAQ: INTC) currently holds a market capitalization of about $485 billion. Nvidia’s projected increase in value alone would therefore equal the creation of roughly 14 companies the size of Intel.

Nvidia > AMD + Micron in Market Value

The comparison underscores Nvidia’s unprecedented scale within the industry. The company is already worth more than Advanced Micro Devices, Inc. (NASDAQ: AMD) and Micron Technology, Inc. (NASDAQ: MU) combined.

• AMD market cap: roughly $778 billion
• Micron market cap: $1.05 trillion
• Combined competitor value: $1.828 trillion

Nvidia remains the first publicly traded company to surpass a $5 trillion valuation. Despite this milestone, analysts argue that demand for its AI infrastructure is still in the early stages.

What the Numbers Show

The divergence between Nvidia’s current valuation and its projected upside indicates that Wall Street expects significant multiple expansion rather than just linear earnings growth. With a current market cap of $5.47 trillion, the implied path to $12.5 trillion requires adding more value than the entire current market capitalizations of AMD and Micron combined. This suggests investors are pricing in sustained dominance across expanding AI use cases, including inference, agentic AI, enterprise computing, and robotics.

During Wednesday’s earnings call, CEO Jensen Huang reiterated that AI factories are expanding beyond model training into these new domains. He expressed confidence that frontier AI companies such as OpenAI and Anthropic will remain Nvidia customers “for a very long time.”