Key Takeaways
Nvidia's next-generation Vera Rubin platform has entered full production, marking the start of a product cycle CEO Jensen Huang says will push the company past $1 trillion in cumulative AI chip sales through 2027.
Nvidia's Vera Rubin platform has entered full production, giving hyperscalers a chip that generates tokens 1.8 times faster than x86 processors and extending the company's dominance into a new $200 billion CPU market.
"This reinvention of the computer is as big of a deal as the reinvention of the phone into what we now know as the smartphone," Huang said during a keynote at Computex in Taiwan on Monday. "Microsoft and Nvidia are going to reinvent the PC. This is the first completely re-engineered, reinvented line of PCs that has happened in 40 years."
A single Vera Rubin system contains close to 2 million components and involves roughly 150 Taiwanese supply chain partners, according to TradingKey. Nvidia is making millions of the CPUs for what Huang called "a market that never existed before." Early customers include Anthropic, OpenAI, SpaceX's xAI, Dell, Oracle and CoreWeave. The company also unveiled the RTX Spark superchip for personal computers, pairing a Blackwell GPU with a custom Arm-based N1X central processing unit designed with MediaTek and built on TSMC's 3-nanometer process.
The production ramp comes as Nvidia targets $1 trillion in combined Blackwell and Rubin chip sales through 2027, double the $500 billion forecast Huang gave at GTC last October. The company is spending roughly $150 billion a year in Taiwan, up from $10 billion to $15 billion four to five years ago, according to Reuters.
Vera Rubin's Architecture Targets the AI Factory Bottleneck
The Vera Rubin platform addresses a specific bottleneck in AI infrastructure: the central processing unit. While graphics processing units handle the parallel math required to train large models, accessing that data and routing it to multiple AI agents requires general compute that only a CPU can provide. Nvidia told CNBC in February that CPUs were "becoming the bottleneck" amid surging agentic AI workflows.
Ian Buck, Nvidia's vice president of hyperscale and high performance computing, said Vera can produce tokens 1.8 times faster than x86 today, "advancing overall agent token performance, enabling smarter, longer-thinking agents and in the end, generating more data center token revenue."
The platform sits at the center of what Huang calls the "AI factory" — data centers designed not just to train models but to run continuous inference workloads for enterprise customers. "Fast CPUs have become essential to keeping the AI factory moving," Buck said.
Nvidia's PC Ambition Puts It in Direct Competition With Intel and AMD
Beyond the data center, Nvidia's RTX Spark superchip marks the company's first serious push into personal computers, a market long dominated by Intel, Advanced Micro Devices, Qualcomm and Apple. The chip combines a Blackwell GPU with 128 gigabytes of unified memory and the Arm-based N1X CPU, and will debut in more than 30 laptops and 10 desktops from Microsoft, Dell, HP, ASUS, Lenovo and MSI starting this fall.
The first laptops will be as thin as 14 millimeters and carry premium price tags, targeting creators, AI developers and gamers. Nvidia said the RTX Spark is "roughly equivalent" to its leading RTX 5070 laptop GPU. The company plans to expand to different price points over time.
The move represents a structural shift in the PC industry. Arm-based processors are gaining ground over the traditional x86 architecture Intel pioneered in the 1970s. Apple already makes Arm-based chips for its MacBooks, and AMD is reportedly working on an Arm-based PC chip of its own.
What It Means for Investors
Nvidia shares have pushed the company's market cap past $5 trillion this year, briefly touching $5.5 trillion in May, according to Stocktwits. Huang told a Taipei audience on May 27 that the company "will be worth even more in three to five years," Reuters reported, and has previously outlined a path to a $10 trillion valuation.
Wall Street is gradually raising its targets. Bank of America has a $320 price target on Nvidia, implying roughly 45 percent upside from current levels. Morgan Stanley analyst Joseph Moore carries an overweight rating with a $260 target and called the current valuation "a surprisingly good entry point" in a recent client note.
The key variable is demand visibility. Hyperscalers are now signing three-year supply contracts, some with full upfront prepayments — a structure that suggests confidence the cycle has not peaked. Nvidia is assuming zero data center compute revenue from China in its second-quarter outlook, meaning any easing of export restrictions would represent pure upside.
This article is for informational purposes only and does not constitute investment advice.
