First, some big picture background: so far, 84.0% of the S&P 500’s market cap has reported. Q1 expectations are for revenues to grow 11.2% and EPS 28.4%. Earnings are beating estimates by 17.3% on aggregate so far, with 78% of companies topping projections to date. The 6 largest megatech companies are expected to outgrow the rest of the market as a group (EPS growth 61.2% vs. 17.9%)

Taking a closer look at Nvidia, it is no secret that this is the most important stock in the world. What it says will color perceptions of the AI boom that has sustained equities and unleashed huge rallies from South Korea to Silicon Valley in the last few months. That could be overshadowed by developments in the Persian Gulf, but on the tech side the next major earnings release won’t be until Oracle’s in three weeks.

Nvidia shares typically fall on earnings, despite the company’s track record of beating estimates and raising forecasts. Investors take profit around earnings, not just in Nvidia, but in tech stocks more generally. And as my noted below, noted, options positioning has the shares primed for a post-earnings dip.

The key number to listen out for in Nvidia’s earnings will be $1 trillion. That’s CEO Jensen Huang’s minimum projection for sales of Blackwell and Rubin AI chips through 2027. Whether, and by how much, Huang is able to raise that projection will determine the direction, not just of its own stocks, but conceivably of stocks worldwide, in coming days and weeks.

The other number to focus on will be the company’s gross margin. Given supply constraints and rising costs in the chip industry and the raw materials pipeline, there’s an obvious risk that rising costs reduces Nvidia’s margins in the future. Forecasts for gross margin in the most recent quarter, and for future quarters, have been stable around 75%, suggesting analysts are confident that Nvidia can push rising costs on to customers.

Investors hoping for a increase or an extension of the $1 trillion forecast are setting themselves up for disappointment, according to Bloomberg. It’s only been two months since Huang made that call at the firm’s GTC conference. Rubin chips in particular include some cutting-edge technology, like HBM4 memory, for which there isn’t a huge production capacity in the world. On the other hand, while supply-chain risks have clearly risen since early March, that was a forecast for demand and orders, not deliveries.

Huang said in March that Vera Rubin systems were already being produced and the first racks were in operation. Production should ramp up in the second half. (Vera Rubin provides multiples more computing power for less energy consumption than previous generations of AI chip.) We’ll be looking for updates on output, as well as timing and forecasts for Groq chips. Nvidia licensed technology from AI startup Groq at the end of last year and Groq chips should start shipping in the second half, probably 3Q.

Should the market reaction to earnings be especially favorable, there is a chance that Nvidia’s market cap could hit $6 trillion tomorrow morning. A summary of sellside consensus for Q1 is shown below:

• Adj EPS 1.774
• Revenue    79.186 BN
• Operating Profit    52.068BN
• EBITDA    52.898BN
• Adj Net Income 43.210BN

And some more detail:

Here’s a quick summary of sell-side versus historical outperformance expectations.

• F1Q (April): Mgmt guided to $78bn in sales, vs. 10 most recent historical outperformance of average 7-8% beat for the reported quarter. This would suggest $83-84bn in sales, vs. current consensus of $79.2bn.
• FQ2 (July): Current consensus is $86.4bn, though 10 most recent guides suggest an average of 3-4% raise, or $89-90bn in guide.

Looking ahead, the key focus during the call and Q&A will have to do with

1. Potential for increased shareholder cash returns,
2. Vera Rubin ramp timing (2H 26E),
3. Gross margin durability (~75% amidst continued memory/other cost inflation),
4. Update to the $1 Trillion 25-27 forecast, esp. contribution from LPU racks, CPU and Vera Rubin Ultra, not included before
5. Potential upside from agentic AI to the server CPU business;
6. Competitive landscape changes against Google TPU, agentic CPU, other ASICs.

Next, we bring you what several banks are watching ahead of the print from the volatility, flow and positioning perspectives, from a handful of banks.

Starting with JPM, we find that while $79BN and 1.77 are the sellside revenue/EPS estimates, the buyside bars are well higher, with the whisper at $81BN and $1.84. Of course, what matters even more is the company’s guidance and here JPM expects a print north of $90BN so the street isn’t disappointed.

Next, UBS trader Joseph Peroni writes that the April quarter likely needs to print $81BN, with July guided to $90–91 BN. Beyond the numbers, upside drivers include capital return (buyback/dividend), confirmation Vera Rubin is on track, updates on central processing unit (CPU) wins tied to Vera, or partnership announcements (e.g., Anthropic).

On positioning, UBS says that Nvidia appears to be used as a funding short (short interest near highs), while hedge fund crowding is near lows and retail has been a heavy net seller over the past month.

* * *

Bank of America (which has a Buy/$320PT rating) has published what is arguably the most detailed preview this time. It starts by flagging flag NVDA’s gross margin durability throughout multiple product cycles, despite a modest setback in early 2025 during initial Blackwell ramp (2H 24 launch, 2025-2026 ramp). Given the upcoming Rubin platform continues to use the same ‘Oberon’ rack architecture as Blackwell and Blackwell Ultra, we expect generally easy product transition in 2026-27 without much GM impact.

Beyond the headlines, BofA expects the focus on:

1. Potential for enhanced cash returns,
2. Vera Rubin ramp timing (2H26E),
3. GM durability (~75% amidst continued memory/other cost inflation),
4. Update to $1Tn CY25-27 forecast esp. contribution from LPU racks, CPU and Vera Rubin Ultra, not included before
5. Competitive landscape changes against Google TPU, agentic CPU, other ASICs.

Even as Rubin Ultra (with new ‘Kyber’ rack architecture) comes along in 2H27 and as HBM continues to represent a bigger part of build cost, BofA flags consensus still models NVDA GM at around ~74% over time, versus current ~75% range.

Next, the bank focuses on how (in its opinion) Nvidia is undervalued based on a valuation discount to the Mag 7. According to analyst Vivek Arya, NVDA is trading at a significant discount to its large-cap growth, Mag 7 peers. On a PE basis, NVDA’s 26x/19x CY26/27E PE is a nearly 50% discount to the Mag-7 average of 49x/42x. On a FCF basis, NVDA’s 28x/20x EV/FCF is an even greater 66% discount. On a normalized Price/Earnings Ratio or PEG, NVDA is trading at a 0.41x PEG, well below peers averaging 2.61x and even the broader S&P 500 market at 1.3x+ PEG

NVDA same FCF as Apple + Microsoft but with 30% less market cap:  The historical pivot towards FCF returns at Apple and Microsoft is well known. While it certainly was not the only reason for their enhanced PE multiples, it likely created a class of sticky shareholders, including possible attraction of dividend growth and income growth funds. While NVDA at $5.46Tn is the largest market cap in the S&P 500 index, it trades at a ~28% discount to the combined $7.5Tn+ market cap of Apple and Microsoft combined despite the potential for generating even greater amount of free cash flow.

NVDA FCF returns have lagged peers: While FCF returns are not the only driver of stock performance, we do believe they expand ownership and signal durability, especially when investors start to get concerned about the sustainability of growth. NVDA’s FCF returns averaged only 47% in the past 3 years (CY22-25), well below peers returning on average 80% and even below NVDA’s own 80% average in the prior decade.

NVDA AI market share outlook. For NVDA’s key AI accelerator business, BofA anticipates it will maintain ~70% market share of the ~$1.2Tn accelerator market through CY30.

In summary, BofA is bullish because NVDA’s product/customer breadth, software/developer support, multi-cloud availability create an integrated, standardized infrastructure that would be difficult for other merchant or custom chip competitors to overtake. BofA expects NVDA to sustain ~70% revenue share in a $1.7Tn+ CY30E AI TAM. Recent noise on CPU/GPU ratio tilting towards CPU in agentic is misleading since: 1) NVDA’s Vera CPU will be a formidable rival in stand-alone CPU, with related newsflow at upcoming Computex, 2) CPU/GPU ratio in widely deployed Blackwell/TPU clusters already at 1:2, and no correlation with agentic CPU in standalone clusters. CPU will be a large but crowded market with x86/ARM option.

* * *

Next, we turn to Goldman’s preview. The bank recaps consensus which as noted above is roughly ~$79bn Q1 revs (although the bullish whisper is closer to >$80-81bn) and Q2 ~$87bn (some estimates, such as UBS >$90bn). Management is expected to guide gross margins to approximately 74.5% to 75.0%. To Goldman trader Nelson Armbrust, a beat alone probably isn’t enough, guidance and gross margins matter more.

Goldman believes that investors will focus on:

• the magnitude of upside to Nvidia’s $1 trillion datacenter guidance at GTC;
• potential upside from agentic AI to the server CPU business;
• competitive dynamics;
• gross margin outlook given rising input costs.

According to Goldman’s desk, NVDA’s positioning score is 9 out of 10 (10 = max long). As the semis complex has rallied in recent weeks, long delta + short gamma exposure via levered ETFs tied to NVDA/semis space has grown with it. The mechanical buy/sell pressure that exists around tonight’s catalyst has potential to exacerbate the stocks’ reaction in both directions as these ETFs (NVDL, NVDX for ex.) need to rebalance daily to maintain their leverage ratios.

What is needed for stock to perform?

1. improving profitability metrics at hyperscalers that supports sustained spending growth;
2. proliferation of agentic AI signaling broader enterprise adoption;
3. more visibility into deployments at non-traditional customers.

Ahead of earnings, Armbrust is in the bearish camp. Thats because “the stock had a good run recently, positioning stretched, long weekend ahead (dont want to own too much risk given geopolitical risk), numbers will be solid and stock will fade as it has on 4 of 5 prints T+1…. NVDA hasn’t really delivered an outsized T+1 move on a print since May’22″

* * *

Speaking of technicals and positioning into the print, SpotGamma writes that the option implied move is ~5.2% (down from ~6.5%). To Spotgamma, the main feature is that implied vol is high – even relative to prior earnings.

As for the latest stats on dealer gamma positioning in NVDA:

1. Well supported below 220 into 200
2. Local negative gamma 230 into 250. Interesting that a 5% implied earnings reaction move puts you at 230.

Also note that the 1m put-call skew for NVDA has turned negative at -0.01, marking the first time it has dipped below zero in over a year. This shows growing demand for upside exposure heading into earnings.

One final point from Bloomberg: while options are pricing in a 5.2% move after earnings, the stock has tended to under-perform the implied, and the earnings -related realized moves have all been under 10% in the last 8 quarters.