Two weeks before Nvidia’s next earnings release, the chipmaker finds itself navigating a paradox. Its biggest hyperscaler customers are pouring record sums into AI infrastructure — much of which lands in Nvidia’s lap — yet one of them, Google, has just opened a direct sales channel for its own in-house silicon.
Google’s TPU Gambit Blurs the Lines Between Partner and Rival
Alphabet announced it will now sell its eighth-generation Tensor Processing Units — the TPU 8t for training and TPU 8i for inference — directly to enterprise customers who want to run them in their own data centers. Until now, Google only offered TPU capacity as a cloud service. CFO Anat Ashkenazi indicated that initial hardware revenue would appear in 2026, though the bigger financial impact is expected in 2027.
For Nvidia, this is no sideshow. Its Data Center segment generated roughly $193.7 billion in fiscal 2026, out of total revenue of nearly $215.9 billion. The unit that powers the AI boom is also the one most exposed to competitive pressure from custom chips.
Nvidia’s counterargument: its own GPUs are programmable across a wide range of workloads, making custom chips a partial rather than full substitute. And Google itself undercuts its own competitive move — the company plans to offer Nvidia’s Vera-Rubin NVL72 systems in its cloud during the second half of 2026, putting TPUs and GPUs side by side.
Hyperscaler Spending Hits Unprecedented Levels
The broader investment backdrop remains Nvidia’s strongest tailwind. Meta raised the upper end of its capex forecast to $145 billion. Microsoft is budgeting $190 billion for calendar 2026 — well above the $154 billion analysts had penciled in. Amazon is targeting $200 billion this year, focused on cloud and generative AI.
Together, Google, Amazon, Microsoft and Meta are planning roughly $725 billion in capital expenditures for 2026, a 77% increase year-over-year. This isn’t just about chatbots anymore. The cloud giants are building capacity for autonomous AI agents, robotics and AI-powered search — applications that demand multiples of current computing power.
Nvidia CEO Jensen Huang said at the GTC conference in March that he sees purchase orders exceeding $1 trillion for Blackwell and the Vera-Rubin platform through at least 2027. A year earlier, he had cited $500 billion. The Vera-Rubin platform, set for volume shipments in the second half of the year, promises to cut GPU requirements for AI training by 75% and reduce inference costs by 90%.
Should investors sell immediately? Or is it worth buying Nvidia?
ServiceNow Deepens Its Nvidia Ties
On the enterprise front, ServiceNow used its Knowledge 2026 conference to significantly expand its six-year partnership with Nvidia. The centerpiece is Project Arc, an autonomous desktop agent for corporate environments. It runs on Nvidia’s OpenShell runtime for security and is managed through ServiceNow’s AI Control Tower. The agent handles complex, multi-step tasks — file access, terminal operations, application control — that traditional automation cannot manage.
Project Arc remains in early preview. The integration of the AI Control Tower with Nvidia’s Enterprise AI Factory design is already generally available. The partnership also includes NOWAI-Bench, an open benchmark suite for enterprise AI agents linked to Nvidia’s NeMo-Gym library.
The May 20 Earnings Test
Nvidia shares trade around €170, roughly 6% below their 52-week high from April. The stock has gained about 70% over the past year — a price that bakes in lofty expectations. On an annualized basis, volatility remains elevated at 34%.
When Nvidia reports fiscal first-quarter 2027 results on May 20, the consensus calls for revenue of $78.8 billion, growth of nearly 79% year-over-year, and earnings per share of $1.77. Management has guided for $78 billion in revenue, implying roughly 77% growth. Analysts are already pricing in 79% growth, and many see 80% or more as the minimum for a positive market reaction. For the second quarter, the Street expects $86.6 billion.
Historically, the stock has more often fallen than risen after earnings — even when headlines beat expectations. The key will be whether management’s guidance exceeds the already ambitious consensus. Nvidia enters the report with net liquidity of $51 billion and a gross margin above 71%.
Analyst targets remain bullish. DBS raised its price target to $250 in late April. Bernstein and Cantor Fitzgerald each see $300, while Rosenblatt targets $325. Whether those levels hold will depend on what Nvidia says about the quarter ahead — and whether the hyperscaler spending spree shows any signs of slowing.
Ad
Nvidia Stock: Buy or Sell?! New Nvidia Analysis from May 8 delivers the answer:
The latest Nvidia figures speak for themselves: Urgent action needed for Nvidia investors. Is it worth buying or should you sell? Find out what to do now in the current free analysis from May 8.
Nvidia: Buy or sell? Read more here...
