NVIDIA is building an AI moat with an unusual density of strategic actions: co-leading the investment in Perplexity, spending $6 billion to acquire Poolside’s team of 100 engineers and software licenses, and making deep investments in data center power infrastructure. Meanwhile, prices for its flagship chips are set to rise by 17% next year, pushing up costs for gigawatt-scale data centers by over $5 billion. Jensen Huang’s strategy is clear: secure ecosystem lock-in through investment ahead of the AI infrastructure boom, and affirm pricing power through price hikes.
NVIDIA is consolidating its core position in the artificial intelligence supply chain with an unusual density of strategic actions. Amid multiple pressures—including rising chip prices, escalating data center construction costs, and increasingly negative public sentiment—Jensen Huang has chosen to accelerate his offensive rather than remain on the sidelines.
According to a Tuesday report by The Information, NVIDIA recently announced that it led a multi-billion-dollar financing round for Perplexity, an AI search and agent company. Simultaneously, the company spent $6 billion to license and acquire certain software assets and talent from AI coding startup Poolside, while continuing to inject capital into developers of data center land and power infrastructure. These three parallel fronts point to a single strategic objective: securing early adoption of NVIDIA’s bundled hardware and software solutions during the initial phase of AI infrastructure development to gain a first-mover advantage.
Meanwhile, server prices for NVIDIA’s flagship Blackwell and Rubin chip systems are projected to rise by approximately 17% next year. Analysts estimate this will increase the construction cost of a gigawatt-scale data center by at least $5 billion. Despite significant cost pressures, NVIDIA’s financial results for the July quarter, released this week, are expected to maintain strong growth momentum.
Data Center Strategy: Securing Early Lock-In Through Investment
NVIDIA’s investment logic regarding data center infrastructure is clear. Jensen Huang has sequentially announced investments in several data center land and power developers in the short term. The aim is to embed NVIDIA’s bundled hardware and software solutions into the data center construction process as early as possible, while also hedging against the risk of chip oversupply caused by issues such as power delivery delays and cost overruns.
Cloud service providers and data center developers are facing unexpected delays and cost increases related to power supply and other factors, creating a window for NVIDIA to intervene early. NVIDIA has made a minority equity investment in Cloverleaf Infrastructure, a company focused on coordinating land, power, and construction resources to provide "ready-to-build" sites for data centers.
Furthermore, NVIDIA ultimately invested in Lancium, the power developer backing OpenAI’s Stargate facility in Abilene, Texas. OpenAI had previously explored the possibility of investing in or acquiring Lancium independently.
Model Strategy: A $6 Billion Bet on Nemotron
At the AI model level, NVIDIA has also demonstrated strong aggressive intent. The company has agreed to pay $6 billion by the end of next year to license certain software from Poolside and recruit 100 employees from the firm. Poolside initially focused on developing AI coding agents but later expanded into the data center and open-weight model sectors.
This transaction underscores Jensen Huang’s commitment to establishing Nemotron as a leading-tier open-source model. This strategic move aims to position NVIDIA not merely as a chip supplier, but as a key participant in the AI model ecosystem, thereby building a deeper moat across the entire AI value chain.
Chip Price Hikes: Cost Pressures Transmit Downstream
NVIDIA's dominant position is also reflected in its pricing power.
Server manufacturers have informed customers that prices for certain flagship NVIDIA chip systems based on Blackwell and Rubin architectures will rise by approximately 17% upon delivery next year. This increase will add at least $5 billion to the construction cost of a gigawatt-scale data center, with cost pressures passing directly to cloud service providers and data center developers.
Nevertheless, these pressures are not expected to appear in NVIDIA's July-quarter earnings report released this week. Jensen Huang and Chief Financial Officer Colette Kress are anticipated to present a series of positive figures to preemptively address external concerns regarding the tightening economic and political environment facing its major chip customers.
External Challenges: Headwinds in Public Opinion and Industry Concerns
NVIDIA's expansion has not been without resistance. Public sentiment in the United States toward AI and the underlying data centers is increasingly negative, and the economic arguments raised by anti-data center activists are difficult to effectively refute. Claims by AI leaders such as Sam Altman, Dario Amodei, and Demis Hassabis that AI will cure cancer have also failed to effectively shift public opinion; indeed, even AI proponents within the pharmaceutical industry are beginning to express resistance toward frontier laboratories.
Meanwhile, the pace of AI commercialization has been slower than expected. OpenAI CEO Sam Altman recently admitted that he underestimated the difficulty of disrupting the software industry with AI, stating that "the economy has too much inertia, and people continue doing the same things." Altman likened the current stage to the period just before the iPhone's launch, suggesting it is largely a "failure at the product level." OpenAI is also facing turmoil within its sales team; following the departure of former Chief Revenue Officer Denise Dresser, Kaylin Voss, the head of sales for the Americas, also resigned recently after only five months in the role.
In contrast, Anthropic is demonstrating stronger growth momentum. The company's banking advisors have indicated to potential investors that its IPO valuation could reach as high as $2 trillion, with fundraising exceeding $100 billion, and that the IPO could launch as early as this autumn.