Reward 500 Tiger-coinsAt the recently concluded GTC 2026, $NVIDIA(NVDA)$ unveiled nearly its entire arsenal: the Vera Rubin architecture pushing the limits of compute, the acquisition of Groq bringing LPUs to strengthen inference capabilities, and the OpenClaw agent strategy. Jensen Huang has effectively completed a transformation—from “selling chips” to becoming a full-stack AI service provider.
Jensen Huang’s $1 trillion outlook briefly pushed NVIDIA’s stock up more than 4.3%.
Yet strangely, the stock has been trading sideways between $170 and $200 for quite some time. Why is Jensen pushing so hard while the market remains so calm?
1. Surrounded by Rivals: Is NVIDIA Starting to Feel the Pressure on Its Monopoly?
To defend its 76% market share, NVIDIA has recently shown signs of urgency.
Strategic investments to lock in customers:It is reportedly investing $30 billion in OpenAI and $10 billion in Anthropic, essentially using capital to secure future orders.
Meanwhile, $Alphabet(GOOG)$’s TPU v7 has already closed the gap to within about one year in FP8 performance. To retain major clients like $Meta Platforms, Inc.(META)$, some of NVIDIA’s recent agreements even appear to include subtle price concessions to lock in long-term demand.
On top of that, $Advanced Micro Devices(AMD)$ has shown strong momentum over the past year, while $Broadcom(AVGO)$’s custom AI ASIC has been booming.
In the coming inference era, where raw performance may no longer be the sole metric, cost efficiency and in-house alternatives could become NVIDIA’s toughest challenges.
2. The Market’s “CapEx Phobia” — Lessons from Meta
Recently, capital markets have shown a fascinating pattern.
When $Meta Platforms, Inc.(META)$ said 2026 CapEx could reach $135 billion (over 50% of revenue), investors reacted with concern. But when reports surfaced that Meta might cut 20% of its workforce to improve AI-driven productivity, the stock jumped nearly 3% pre-market on Monday.
This reveals the market’s true mindset: Investors are starting to question how long Big Tech can keep pouring money into AI chips.
When CapEx reaches 50% of revenue, unless downstream applications (like Meta’s ad system or AI agents) produce clear monetary returns, that level of spending may not be sustainable.
3. NVIDIA’s “Second Derivative” Problem
NVIDIA sits at the second derivative of spending.
This means that even if hyperscalers like Meta or Google keep CapEx high, once the growth rate slows, NVIDIA’s revenue from these clients could effectively stall.
So while Jensen emphasized a “$1 trillion total opportunity” at GTC, many analysts view that number as in line with expectations—or even slightly below some aggressive forecasts.
💬 Discussion
Meta’s potential large layoffs—are they:
-
A sign of AI-driven productivity gains, or
-
A signal that Big Tech must cut costs to preserve profits?
Ultimately, whether CapEx in 2025–2026 continues to exceed expectations may determine if NVIDIA can break through the $200 ceiling.
A. Bullish: AI agents will create another compute shortage.
B. Cautious: Big Tech CapEx has peaked; the next phase is efficiency and consolidation.
C. Wait and see: Let’s watch Big Tech earnings guidance in May before deciding.
Drop your thoughts in the comments—how do you view this wave of “cost cutting + AI efficiency” across Big Tech?
Comments
The bigger concern is demand quality. When $Meta Platforms, Inc.(META)$ and $Alphabet(GOOGL)$ keep spending but shift toward efficiency, it suggests CapEx growth may be peaking. For NVIDIA as a second-derivative play, that matters more than absolute spending, especially with pressure from $Advanced Micro Devices(AMD)$ and $Broadcom(AVGO)$ .
So I’m not bearish—just patient. I still believe in AI demand, but for NVIDIA to break higher, we likely need either re-accelerating CapEx or clearer AI-driven revenue. I’ll wait for the next earnings cycle before getting more aggressive.
@TigerStars @Tiger_comments @TigerClub
Meanwhile, $Alphabet(GOOG)$’s TPU v7 has already closed the gap to within about one year in FP8 performance. To retain major clients like $Meta Platforms, Inc.(META)$, some of NVIDIA’s recent agreements even appear to include subtle price concessions to lock in long-term demand.
The narrative that Big Tech CapEx has peaked is premature. We are moving from the "Training Phase" (building models) to the "Inference Phase" (powering AI Agents). As AI Agents become integrated into every WhatsApp message, Instagram ad, and enterprise workflow, the sheer volume of continuous, real-time computation required will trigger another massive compute shortage. Meta, Microsoft, and Google cannot afford to stop buying H200s or Blackwells because falling behind in AI agency is an existential threat. These layoffs are the "fuel" being rerouted to power the engines provided by NVIDIA.
In short, "cost cutting" is the tactical move to fund the strategic expansion of AI. As long as the ROI on AI-driven ad targeting and automated coding remains visible on the balance sheet, the CapEx faucet will stay wide open through 2026.
My Take: B. Cautious (with a twist)
I lean toward B, but with a caveat: The "peak" might be a plateau rather than a drop. We are moving from the Build Phase (buying every chip available) to the Deployment Phase (making those chips pay for themselves).
NVIDIA is trading sideways because the market is waiting for a "killer app" (like the OpenClaw agents) to generate enough revenue for Meta/Google/Microsoft that they have to order another $100B in Vera Rubin chips.