Nvidia projects 70% growth and signals that the AI investment cycle will last longer

Nvidia signaled this week that the cycle of investment in AI infrastructure is far from cooling off, and the market reacted immediately. According to Reuters, via Economic Times, the company's shares rose 6.7% in Thursday's pre-market trading, to US$223.71, putting the company on track to add roughly US$340 billion to its market value.

What moved investors was the forecast: Nvidia estimates revenue growth of 70% in the next fiscal year and expects current-quarter sales above Wall Street estimates. It's an explicit signal that global AI spending, the famous AI spending boom, remains intact.

What came before

The optimism wasn't guaranteed. Nvidia's shares had fallen nearly 12% from their May peak, amid growing pressure from investors for concrete proof that the AI spending boom would hold up. The underlying question was classic for hype cycles: would hyperscalers keep buying GPUs at the same pace, or had the market inflated beyond what real demand justified?

The long-term forecast was, according to the report, a rare statement from the company about an extended horizon. CEO Jensen Huang said AI had reached an "inflection point," with the technology moving from the experimentation phase to production deployment. In other words: demand would no longer be driven by tests and proofs of concept, and would instead be pulled by systems actually running in the real world, which tends to be more predictable and recurring.

The quarter's numbers

Nvidia reported revenue of US$96.2 billion in the second quarter, above expectations, driven by US$89 billion in data center sales, the core of the AI business. After the results, at least 10 brokerages raised their price targets for the stock, according to data compiled by LSEG.

Morgan Stanley analysts summed up the astonishment: "70% growth while supply constrained is a remarkable number, and, as much as possible, we expect Nvidia to keep breaking through barriers to grow even further." The "supply constrained" detail matters: the company is growing even without being able to meet all demand, which means supply bottlenecks remain on the radar.

A point of context for those who follow valuation: the stock trades at 17.9 times estimated earnings, well below AMD's 37.2 and Intel's 46.2.

Demand is spreading

A relevant signal for those building software is that Nvidia indicated AI demand is going beyond the major hyperscalers. The company cited growth coming from AI labs, increased capacity among neo-cloud providers like CoreWeave and Nebius, and a deeper partnership with Amazon Web Services.

CoreWeave and Nebius shares rose 5.8% and 7.2%, respectively. Morgan Stanley also noted that Nvidia's move into cloud revenue-sharing models could become a new catalyst for the stock, a shift that brings the chipmaker closer to the business of selling computing capacity, and not just hardware.

What this changes for developers in Brazil

For the Brazilian developer, the practical effect isn't in the stock price, but in what it signals about the coming years of infrastructure. A few points worth watching:

• Cloud and GPU costs remain under pressure. With Nvidia growing even while constrained by supply, top-tier GPUs will remain scarce and expensive. Anyone running inference or training in the cloud should expect to keep dealing with queues, quotas, and prices that don't fall at the pace one would expect from mature hardware.
• More options beyond the three giants. The growth of neo-cloud providers like CoreWeave and Nebius widens the range of places to rent GPU capacity. For teams that currently depend only on AWS, Azure, or Google Cloud, these players could become a cost alternative, although their presence in Latin America, and the latency to reach it, need to be assessed on a case-by-case basis.
• Roadmap planning with a longer horizon. The message that the investment cycle will extend for years gives more predictability to those deciding how to architect AI-backed products. Betting on inference pipelines, RAG, and agents is likely to find an infrastructure ecosystem that's expanding, not shrinking.
• The phase has shifted from experimentation to production. If Huang is right about the "inflection point," the day-to-day bottleneck stops being "getting a GPU to test" and becomes reliably operating AI workloads with controlled costs in production.

What remains open

The report itself notes that enthusiasts still live with two concerns: chip supply bottlenecks and Nvidia's financial ties to its customers, a sensitive point when the same company sells hardware, gets closer to cloud providers, and starts sharing revenue with them. There's also no guarantee, within the forecast, that hyperscalers will keep up their buying pace indefinitely; the 70% growth figure is a company estimate, not a consolidated data point.

For the Brazilian market, what's left is to watch whether and how this extended cycle translates into GPU capacity actually available in the region, and at competitive prices. For now, the message is one of continuity: the infrastructure underpinning the AI wave should keep growing, and the architecture and cost decisions made now will have to coexist with this scenario for quite a while.