Nvidia Stock Price: (NVDA) $167 Is the Opportunity — Three Quarters of Acceleration and a 20x P/E Below the Sector Median

Nvidia (NVDA) is trading at $167.64 Friday, off 2.1% on the session and 17% below its $212.19 52-week high. The stock has traded essentially sideways for six months. Meanwhile, Q4 FY2026 delivered record revenue of $68.13 billion — up 73.21% year-over-year — EPS of $1.62 up 82%, net income of $42.96 billion with a 63% net margin, and return on capital of 74.22%. Cash sits at $62.56 billion. Free cash flow hit $14.69 billion in the quarter, up 49.21%. Full-year FY2026 revenue was $216 billion, up 65%. The market's response to all of this has been to price NVDA at 20.41x forward earnings — below the IT sector median. That disconnection between one of the greatest operational performances in technology history and a sub-sector-median valuation multiple is the entire investment case.

Three Consecutive Quarters of Revenue Acceleration — The Inflection Is Confirmed

The direction of NVDA's growth rate matters more than the absolute number. After an extended deceleration phase that worried the market through much of 2025, revenue growth has now accelerated for two straight quarters. Q4 came in at 73.21% year-over-year. Q1 FY2027 guidance of $78 billion at the midpoint implies 77% growth — the third consecutive quarter of acceleration. That guidance came in above consensus, meaning even analysts who follow NVDA closely underestimated demand. Sequential acceleration at $68 billion quarterly revenue requires genuine structural demand expansion, not product cycle timing. CEO Jensen Huang's framing is precise: "Compute equals revenues. Without compute, there's no way to generate tokens. Without tokens, there's no way to grow revenues." When hyperscalers can directly monetize every GPU they install through token generation, the traditional semiconductor cyclicality argument weakens structurally.

Data Center at $62.3 Billion — But the Real Story Is Networking Up 267%

NVDA's data center segment generated $62.3 billion in Q4, up 75% year-over-year, outpacing even the company-wide growth rate. GPU compute revenues grew 57% — impressive at this base. But networking revenue surged 267% year-over-year. Nvidia now claims the world's largest networking business. The strategic significance is that networking deepens the switching cost moat beyond GPUs alone. When a hyperscaler builds AI infrastructure using both Nvidia GPUs and Nvidia networking simultaneously — InfiniBand, Spectrum-X, ConnectX — replacing GPUs with a competitor's hardware requires simultaneously replacing the entire network fabric. The GB300 NVL72 delivers 35x lower cost per token than the Hopper platform and 50x more tokens per megawatt. When the price per token drops 97% while system prices hold, hyperscalers have no economic incentive to switch to alternatives. Gross margins expanded 1.7 percentage points year-over-year despite the competitive noise — proof the moat is intact.

Q1 Guidance: $78 Billion Revenue, 75% Gross Margin — Three Straight Acceleration Quarters

Q1 FY2027 guidance at $78 billion midpoint represents 77% year-over-year growth. Gross margin guided at 75% — improving from 71.3% year-ago. Operating expenses are projected at $7.5 billion, implying 109% year-over-year growth — the one number that gives bears ammunition. But 109% opex growth against 77% revenue growth at this scale is a deliberate investment phase, not structural deterioration. NVDA executed the same pattern before every prior product cycle acceleration. Huang stated that the company sees "$1 trillion plus of Blackwell plus Rubin" demand visibility through 2027. Wall Street consensus for FY2027 and FY2028 combined is approximately $850 billion — meaning the CEO's visibility statement exceeds even analyst consensus. That gap between what management sees and what analysts model is where upside surprises get manufactured.

Agentic AI Is the Demand Durability Argument — And It's Backed by Numbers

The bear case on NVDA centers on hyperscaler "sufficiency" — the point where new GPU orders become incremental rather than ongoing. The bull case rests on agentic AI as a structurally larger demand category than chatbots or generative media. Klarna's AI assistant reduced customer query resolution time from 11 minutes to 2 minutes, cut repeat queries 25%, and is displacing the equivalent of 700 employees. That is direct, measurable labor cost substitution. As more enterprises deploy agents at scale, compute demand transitions from discretionary capex to operational necessity — the same shift that made cloud infrastructure irreplaceable. CFO Colette Kress: "Adoption is skyrocketing, and tokens are profitable, driving extreme urgency to scale up compute." Inference workloads accounted for 50% of all AI compute in 2025 and are projected at two-thirds in 2026. Inference favors NVDA's GPU versatility over competitor ASICs across diverse workloads. Amazon committed to purchasing another 1 million NVDA chips by end of 2027. Elon Musk confirmed SpaceX, xAI, and Tesla will continue purchasing at scale. The top five hyperscalers still represent 60% of NVDA's revenue despite simultaneously building their own chips — the most credible possible validation of GPU indispensability.

China H200 Approved, Wells Fargo Estimates $25B Additional Annual Revenue

The single most underappreciated near-term catalyst is China. On March 17, Huang confirmed orders from China and manufacturing restart. One day later, the Chinese government approved the H200 for sale. Wells Fargo estimates China could add $25 billion or more to NVDA's annual revenue — approximately 12% of projected FY2027 revenue on top of existing guidance. China has been embedded in NVDA's valuation as a headwind for over a year. Its reversal to a tailwind — even partial — represents a meaningful earnings catalyst not reflected in the current 20x forward multiple. The Groq licensing deal adds inference-optimized architecture to the China revenue layer. Antitrust scrutiny from two senators represents noise, not enforcement risk — licensing structures of this type are standard in semiconductors.

Space-1 Vera Rubin Module: The Next Frontier That Has No Competitors Yet

NVDA released the Space-1 Vera Rubin Module at GTC 2026, targeting SpaceX's planned space data center buildout. SpaceX is planning an IPO specifically to raise capital for space-based compute infrastructure. The addressable market is unquantifiable today, but NVDA's move to position Rubin hardware in a market with zero established competitors and SpaceX as the anchor customer is exactly the kind of first-mover positioning that compounds over years. Space compute is not a 2026 revenue story — it is a 2028-2030 story that will matter when the market eventually re-rates NVDA for duration rather than discounting it for cyclicality.

The Competition Risk Is Real — But 60% of Revenue Still Comes From Hyperscalers Building Their Own Chips

Google, Amazon, and Meta are all building custom ASICs. Wedbush called them "aggressively focused" on alternatives. This is a legitimate risk. But the fact that these same companies — who have the engineering talent and capital to build world-class silicon — still collectively represent 60% of NVDA's revenue is the most honest assessment of GPU competitive moats available. ASICs optimize for specific known inference tasks. GPUs handle the full spectrum of training, diverse inference, and research workloads that no enterprise without a dedicated chip team can replace with custom silicon. The CUDA software ecosystem — the oldest and most comprehensive GPU-optimized AI software library in existence — creates switching costs that persist even if competing hardware eventually matches GPU performance on specific benchmarks. AMD is the most credible challenger, but NVDA's lead in both hardware performance and software ecosystem means AMD gains share at the margin, not in a displacement scenario.

Valuation: 20x Forward P/E Below the IT Sector Median, 0.54x PEG, FY2028 at 15x

The valuation arithmetic is clean. Forward P/E of 20.41x is below the IT sector median — meaning NVDA is cheaper than the average IT stock on forward earnings despite growing at 73% with expanding margins. PEG ratio of 0.54x is less than half the IT sector median of approximately 1.2x. The multiple compresses rapidly: at current price and FY2027 consensus EPS of $8.30, the multiple is 20.2x. At FY2028 consensus EPS of $11.10, it compresses to 15.1x. A 24x conservative steady-state multiple applied to FY2028 EPS of $11.10 produces a price target of approximately $266 — 58% upside from $167.64. Mid-teens total returns through 2030 under this framework require no heroic assumptions about China or space or agentic explosion. They require only that the current trajectory continues. For the risk-optimized approach: selling the July 17th $160 put at $10.48 generates 7% cash-on-cash in 111 days, annualizing to 23%. If assigned, effective entry is $149.52 — 19.3x forward earnings — the most attractive NVDA valuation available in the current cycle.

The Verdict: Buy — The Market Is Discounting Durability That the Numbers Confirm

NVDA at $167.64 is a buy. Six months of sideways price action against six months of accelerating fundamentals, expanding margins, and $1 trillion in CEO demand visibility has created the widest gap between operational reality and market price that NVDA has offered since the AI buildout began. The agentic demand thesis is not speculative — it is visible in Klarna's 700-employee displacement, in Amazon's 1 million chip commitment, in three consecutive quarters of revenue acceleration. China reopening adds $25 billion in potential annual revenue not in consensus. The 20x forward P/E below sector median on a business with 74% return on capital and 63% net margins is not a valuation that reflects the fundamentals. Buy outright with a stop below $150. Sell July $160 puts at $10.48 for 23% annualized yield as the risk-optimized alternative.

That's TradingNEWS