Scoreboard
Hey there, bargain hunter. Nvidia closed the week at $219.22, a 10.47% five-day gain that lifted the stock to its highest level since early June. The week started with renewed AI capital expenditure signals from the hyperscaler complex, accelerated Tuesday on the SpaceX exclusivity announcement, and held gains even after a Thursday report that Nvidia is considering memory configuration cuts on its next flagship chip. Two steps forward, one wobble. Net result: the stock is moving again.
Shares remain well below the record closing high, leaving the 2026 performance notably more muted than the outsized gains investors grew accustomed to during 2023, 2024, and 2025. That gap is not random. It is the market doing arithmetic on a company that has already delivered $215.9 billion in annual revenue and is now being asked to do it again at a bigger number. The next test is August 26, when Nvidia reports Q2 FY2027 results covering the fiscal quarter ending July 26, 2026, and the question is whether revenue met or exceeded the company’s own guidance of approximately $91 billion.
The Real Reason the Week Happened
This was not a random technical bounce. Three events converged.
First: SpaceX. SpaceX and Nvidia are jointly developing the Starmind AI1 satellite compute payload, an orbital compute node equipped with Rubin GPUs and Vera CPUs. The partnership was discussed alongside SpaceX’s first post-IPO earnings report, which showed revenue up 92% year-over-year. More important than the satellite itself was the word Elon Musk used on the call. Musk told investors SpaceX will build its AI infrastructure exclusively on Nvidia hardware, making Nvidia the sole AI accelerator supplier for both terrestrial data centers and planned orbital compute programs. That word, exclusively, is doing significant work. It eliminates AMD from a fast-growing customer’s budget indefinitely.
Second: the capex cascade. Alphabet recently increased its 2026 capex guidance to a new range of $195 billion to $205 billion. Amazon has reiterated that it plans to invest roughly $200 billion in capex in 2026. Both moves signal that hyperscalers are not pulling back from AI infrastructure, they are accelerating. Every dollar of their capex is, to a large degree, a dollar that eventually converts to Nvidia hardware revenue.
Third: the prior week’s fear was simply wrong. Semis took a bruising in the prior week over fears about the ongoing AI buildout and whether the companies’ lofty valuations were supported by fundamentals, but they recovered as traders turned more positive on the sector. When the fear is macro and the data are company-specific, corrections like that week tend to be short-lived.
Deep Dive: What Nvidia Actually Is in 2026
Most investors still instinctively describe Nvidia as a chip company. It is not. It is an AI infrastructure platform that happens to manufacture the hardware at the center of it.
Nvidia remains the dominant AI compute supplier, and its data center business is the primary growth engine, driven by demand for AI training and inference accelerators from cloud providers, enterprises, and sovereign nations building AI infrastructure.
The current product generation is Vera Rubin. The first Starmind satellite compute payload will feature Nvidia’s Vera CPUs and Rubin GPUs, as well as a Vera Rubin NVL72-class system. On the ground, the NVL72 is already the reference architecture for large-scale AI deployments. The design focus is not just raw performance, it is serviceability and speed of deployment, including a cable-less compute tray approach that cuts assembly time dramatically versus prior systems. That is not just a performance advantage. It is a total-cost-of-ownership advantage that makes competing architectures harder to justify on a spreadsheet.
Despite facing intensifying competition from AMD, Intel, and custom silicon solutions from hyperscalers, Nvidia’s CUDA ecosystem and software moat provide significant competitive advantages that extend beyond hardware specifications. The software lock is the part of the story that rarely makes headlines and is the part that matters most to durability.
The sovereign AI business deserves its own line. Sovereign AI revenue was over $30 billion in fiscal 2026, more than tripling year-over-year. Governments are not buying compute for efficiency. They are buying it for strategic independence. That category does not shrink when the next product cycle arrives. It grows.
Data Section
Here is what the business looks like on the numbers heading into August 26:
- FY2026 revenue: Record Q4 revenue of $68.1 billion, up 20% sequentially and 73% year-over-year. Full-year FY2026 revenue reached $215.9 billion, up 65% from the prior year.
- Q1 FY2027 (most recent quarter): Revenue jumped to $81.6 billion, up 20% from the prior quarter and 85% compared with the same period in 2025. The data center business was the main driver, with quarterly revenue surging 92% year-over-year to $75.2 billion.
- Q2 FY2027 guidance: Management guided fiscal Q2 revenue to $91 billion, plus or minus 2%, with gross margins of approximately 75%.
- Full-year FY2027 consensus: Analysts have set full-year revenue estimates in the high $300 billions range.
- Profitability: Return on equity is 114.29%.
- Balance sheet: The company carries a debt-to-equity ratio around 0.07.
- Shareholder returns: During fiscal 2026, Nvidia returned $41.1 billion through share repurchases and dividends. As of the end of Q4, the company still had $58.5 billion remaining in its share repurchase authorization.
Is It Cheap?
Here is the honest answer: it depends entirely on whether you believe FY2027 estimates are real.
Valuation multiples are moving targets week to week, but the setup is simple: the market is willing to pay up if earnings keep outrunning expectations. Whether you trust the denominator is the whole debate.
For comparison, AMD still trades at a materially higher forward earnings multiple than Nvidia. Nvidia’s multiple compresses faster than its revenue because the earnings growth keeps outrunning the stock. That is a good problem to have.
Analyst price targets and ratings also move quickly, but directionally the Street remains broadly positive heading into August 26.
Bull / Base / Bear
Bull case: August 26 delivers revenue above $91 billion, management raises Q3 guidance, and the Rubin Ultra supply risk is resolved before it affects shipments. The SpaceX orbital compute program becomes a proof point that Nvidia’s hardware extends beyond terrestrial data centers. Target range: $250 to $300, consistent with the majority of analyst price targets.
Base case: Q2 revenue lands at $91 billion to $94 billion, gross margins hold near 75%, and Q3 guidance comes in modestly higher. Rubin Ultra ships with reduced memory but customers absorb it because the alternative is waiting. AI chip capacity for 2026 is widely viewed as constrained, so customers have limited options. Stock grinds toward the prior all-time high over the following two to three months.
Bear case: The HBM issue gets worse, not better. Reports suggest Nvidia may ship a Rubin Ultra variant with lower on-package memory than initially expected due to supply constraints and multiple HBM configurations under evaluation. If that downgrade is severe enough to slow model training performance per chip, hyperscalers could defer orders while they evaluate alternatives. Additionally, TrendForce has projected much faster shipment growth for custom AI ASICs than for merchant GPUs in 2026. That gap is narrowing over time.
Action Plan
The earnings date is August 26, eighteen days away. That creates a specific decision framework.
If you own no NVDA: the SpaceX exclusive announcement and the hyperscaler capex escalations are not noise. They are multi-year demand signals. A position initiated now, ahead of August 26, carries event risk on both sides. A scaled entry, half now and half on any pullback toward $210, manages the event risk without surrendering the thesis.
If you own NVDA and are sitting on a gain: trimming here requires a specific bear thesis beyond general market caution. Hold is the default.
If you own NVDA and are sitting on a loss from the prior high: the gap between current price and that peak is single digits. The business has grown substantially since May. Patience is the right position.
Cheap Investor Scorecard
- Revenue growth: 85% year-over-year last quarter. Q2 guided to $91B. Watch for a beat above $93B on August 26. Target: green.
- Gross margin: Approximately 75% guided for Q2 FY2027. Any reading below 74% would be a yellow flag. Watch closely.
- Data center revenue share: In Q1 FY2027, data center revenue was $75.2B of $81.6B total revenue, roughly 92% of the quarter. That concentration is a risk and a strength simultaneously.
- Rubin Ultra HBM specification: The final memory configuration has not been confirmed. Any clarity on this before or during August 26 is a key data point.
- Q3 revenue guidance: Street expectations are clustered in the mid-$90 billions. A guide below $93B would disappoint. A guide above $97B would re-rate the stock.
- SpaceX Starmind timeline: SpaceX has pointed to launches beginning in 2027 for the Starmind AI payload concept. Any acceleration or delay in that schedule changes the near-term revenue math for the orbital compute category.
- Sovereign AI revenue: Over $30 billion in FY2026. Watch for any commentary on FY2027 trajectory.
- Forward PE vs. growth: If the multiple expands meaningfully on August 26, it will be because the guide forces FY2027 estimates higher, not because the market got generous.
- HBM price trajectory: The memory shortage gives suppliers unusual leverage and suggests that elevated HBM pricing may persist. Rising input costs compress margins. Watch Nvidia’s cost-of-revenue line on August 26 against the gross margin guide.
- AMD competitive response: AMD’s rack-scale push is the competitive variable worth tracking into Q4.
Bottom Line
The week’s 10% move was earned, not random. SpaceX formalized a multi-year exclusive demand signal. Hyperscalers confirmed the capex cycle is still accelerating. And the stock, despite all of it, still sits below its May peak with a valuation that is no longer obviously stretched relative to its growth.
The thesis is straightforward: if Nvidia reports $91 billion or more on August 26 and guides Q3 convincingly higher, the distance to a new high closes fast. If the HBM situation produces either a gross margin miss or a weaker guide, the stock can retrace to the $200 to $210 range.
The Rubin Ultra memory downgrade story is the one bear argument worth monitoring seriously. It is a supply failure with design consequences. DRAM scarcity has reached the threshold where it is influencing the specification of the world’s most anticipated AI chip, not merely constraining the price at which that chip is sold. That is a real structural risk, not a headline risk. Keep it on the list.
Everything else argues for owning NVDA into August 26 with a clear exit plan if the data center revenue growth rate decelerates materially. It has not decelerated yet. The hyperscalers are spending more, not less. And this week, SpaceX put its stake in the ground: Nvidia, exclusively.
