#英伟达财报周 NVIDIA's explosive earnings: Can they drive the semiconductor industry chain?

After the U.S. market close, NVIDIA delivered blockbuster results for the second quarter of fiscal 2027: total revenue reached $96.2 billion, up 106% year over year; data center revenue surged 117% year over year to $89 billion, hitting another record high. NVIDIA's earnings were nothing short of explosive.
These earnings are not merely a celebration of one company's performance, but also a “hard validation” of the entire AI semiconductor industry chain—AI has moved beyond the laboratory-concept stage and officially entered an era of monetization with real money, while insufficient supply has become the biggest constraint on the entire industry. At least through fiscal 2028, the upstream and downstream chip and semiconductor sectors as a whole will remain in a strong boom cycle characterized by demand exceeding supply. Jensen Huang said on the earnings call that artificial intelligence has crossed a historic inflection point. In the past, AI largely remained in laboratories and proof-of-concept stages; today, it has become a tool capable of generating productivity and commercial profits. Even more above market expectations was the guidance: NVIDIA expects fiscal 2028 revenue to grow approximately 70% year over year, far above the market's previous consensus estimate of 45%. But even this 70% does not represent true market demand.
Management disclosed that if supply-chain constraints were excluded—in other words, if the company were not limited by insufficient capacity—its revenue growth could be even higher, with potential customer demand approaching a doubling. In a word, demand exceeds supply. Many people previously worried that the AI rally was based on “concept speculation without realized revenue,” but NVIDIA's earnings, backed by solid financial data, have dispelled that concern.
The five leading cloud providers' capital expenditures will approach $800 billion in 2026 and climb to $1.3 trillion in 2027, while the cloud industry's total order backlog has surpassed $2 trillion. Compute deployments are directly driving simultaneous expansion in cloud providers' revenue and profit margins. Regional AI compute factories are being built around the world—in Armenia, Africa, India, Australia, Malaysia, Japan, and Europe, NVIDIA AI supercomputers are being deployed everywhere.
Global AI venture investment exceeded $400 billion in the first half of 2026, with 70% of the funds used to purchase compute hardware. Drug development, automobiles, financial quant, manufacturing, healthcare, and every other sector are integrating AI into their business processes. Compute is no longer exclusive to technology companies, but a critical need for the entire economy. Demand is firing on all cylinders, but the industry's pain points have also been fully exposed. Memory chips are the most immediate issue. NVIDIA stated bluntly that the global memory shortage has not eased, memory prices are far above expectations and will continue rising next year, and higher component costs will directly put temporary pressure on gross margins, with Q4 gross margins expected to fall to 71%-72%. This was also the logic behind the broad rally in U.S. memory stocks after the earnings release, with Micron, Seagate, and Western Digital all rising. AI servers' consumption of HBM and DRAM is exploding. Memory is no longer a simple supporting component, but has become a critical bottleneck constraining compute expansion. Memory manufacturers are closely partnering with NVIDIA to expand production and meet demand for the next-generation Vera Rubin platform. Next is the optical networking and communications sector. Spectrum‑X Ethernet revenue grew 2.6 times year over year, and NVIDIA has grown into a leading global high-speed networking provider. The new-generation Vera Rubin platform's complete systems have even greater bandwidth requirements for high-speed interconnects and optical modules, and optical communications companies such as Lumentum and Coherent also rose in after-hours trading. In addition, NVIDIA has officially entered the server CPU market. Vera CPU is already in volume production and delivery, with specialized optimizations for agentic workloads. Its CPU business is targeted to more than double in fiscal 2028, officially placing NVIDIA among the world's leading server CPU suppliers.
Returning to the core question: Can NVIDIA's explosive earnings really drive the entire semiconductor industry chain?
The inflection point for AI commercialization has arrived. Agent-driven inference demand is surging, with cloud providers, enterprises, sovereign clouds, and AI startups all pushing forward simultaneously. Capital expenditures continue to increase, and real demand far exceeds current supply capacity. At least before fiscal 2028, the upstream and downstream semiconductor sectors related to compute will broadly operate in an environment of demand exceeding supply. Memory, high-speed optical communications, server CPUs, high-speed interconnects, complete systems, and advanced packaging will all directly benefit from the large-scale construction of AI factories worldwide. But bottlenecks exist throughout the supply chain: expanding foundry capacity, HBM memory capacity, and high-speed optical chip capacity all require long lead times and cannot be released in the short term. At the same time, intensifying competition, rising costs, and temporary pressure on gross margins are all realities the industry chain must face. Concept speculation will be disproven, but the hard-core segments capable of delivering products and becoming integrated into global AI compute construction will continue to benefit from this AI supercycle.
NVIDIA's earnings are not simply positive news for one company, but an official endorsement of the prosperity of the global AI semiconductor industry chain. AI is moving from storytelling to spending money on hardware and generating returns through real-world deployment. The great era of compute semiconductors is far from over.$NVDA ‌