NVIDIA’s Earnings Just Changed My Price Target

Last night, Nvidia produced the kind of quarter that would look fictional for almost any other company. Nearly $100 billion of revenue in just 3 months. Revenue that more than doubled. Earnings per share that more than doubled. And the company still guided above Wall Street's expectation for the next quarter. And before the opening bell today, Nvidia's up more than 6%. and semiconductors. They also look to be starting well, even while much of the S&P we can see across the board looks to be sitting in the red. But here's the problem. A brilliant company is not automatically a brilliant purchase, especially when the market already knows it's brilliant. And the report, it also contained lower free cash flow, a soft gross margin guide, and questions about circular financing, and a debate over whether today's demand can survive into 2029. So, similar to Wall Street who have an average price target as well as both a bare and bull case, I've done exactly the same. Rebuilt the valuation based on this latest data and the spread between them's almost $100 per share and my previous target that's now changed. And stay with me because by the end of the episode, I'm going to give you a new target, the assumptions required to reach it, and the price at which this thesis that starts to break. And before we analyze the numbers, one reaction captured the strange standard Nvidia now faces better than any headline. Listen to how quickly impressive became not impressive enough. >> I think my initial impression is that's a a really impressive quarter and nobody's going to care. >> Yeah. >> Not impressive enough, I guess. >> Yeah. I mean, not impressive enough, but it gets to this idea, and I think I asked this question of a lot of our guests leading up to this is what can Jensen Wong and company actually say that does impress this market. It's not in the earnings release, but maybe it comes on the conference call. That skepticism matters when perfection is expected. Merely beating estimates is not the catalyst. The catalyst has to be evidence that exceptional growth lasts longer than investors currently believe. And look, Nvidia actually trailed the boom in several smaller chip makers this year. the company was delivering. The stock was waiting for proof that the duration had extended. And we can see the performance of the market in just the last few months. It's why today's move matters, but it's also while not treat one pre-market rally as confirmation that the valuation questions disappeared. So look, let's start with the clean scorecard. Second quarter fiscal 27 revenue that was 96.2 billion against around 92.2 billion expected. That is a 4 billion beat and adjusted EPS that came in $222 against $29 revenues we said that was up double digit 106% EPS that grew 111 but also this is not growth bought with collapsing profitability. Nvidia's trailing gross margin remained above 74% while both EBIT and in fact net income margins well they remain extraordinary by any industry standard. and compare Nvidia's revenues. We can see 96.2 billion against the previous quarter of 46.7 billion. While they added almost $50 billion of quarterly revenue, I mean very few public companies generate $50 billion in a year. Nvidia added it in just one. And then when we look at their net income, well, it reached just shy of $60 billion. It means the company converted roughly 62 cents of every revenue dollar into net income in just this quarter. And we have the next quarter revenue guide that's at $108 billion plus or minus 2%. Look at what Wall Street estimating $ 105 billion. So that was very nice to see them coming above. On top of that gross margin guided around 74% plus or minus half point. Yeah, I would argue that's very elite but it is slightly lower than what consent has had at 74.8. 8 also slightly lower than yesterday's release of 75% and operating expenses rose much more slowly than revenue allowing adjusted operating income to reach $64 billion. Operating leverage is why a small revenue beat can actually create a larger earnings beat and Nvidia also returned roughly $26 billion through both dividends and repurchase. The dividend, it remains tiny, but the repurchases help offset dilution and return surplus cash while the core business compounds. And something to point out here, the $108 billion guide that actually excludes data center compute revenue from China makes the guy cleaner, but it also leaves geographic upside unavailable unless export conditions and product approvals change. So, look, overall, the headline verdict here is simple. Revenue, earnings, and guidance. The only obvious yellow flags I'd say was cash conversion and the direction of the gross margin. Now the longerterm chart makes this clearer. Nvidia's quarterly revenue has gone from roughly $1 billion to more than 96 billion. The curve is no longer merely exponential looking is almost vertical. I mean if we look at just the last four years quarterly revenue was around $5.9 billion. Today it's more than 16 times larger. This is a mega cap behaving like an early stage growth company. And just on an average calendar day, Nvidium now generates more than $1 billion of revenue. That includes weekends, holidays, mornings when we're asleep every single day. And even more surprising is that growth accelerated as the base became larger. Just look at the same quarter last year. Well, in fact, we can see revenue growth was up 56%. Now, take a look at it. It's up 106%. This is absolutely extraordinary and normally the law of large numbers forces growth down. Nvidia's temporarily reversed that law because the industry is moving from experiments into fullscale AI infrastructure deployment. And Jensen hangs as we can see here, he described the moment as an inflection point and said in effect compute is no longer simply a cost center. Compute itself is becoming revenue. And this phrase matters because it essentially explains why customers can keep spending if additional compute directly creates products, agents, and revenue. Demand is constrained by return on investment rather than by a fixed IT budget. And look, data center revenue reached $89 billion. That's up $17% year-over-year. This single segment is now larger than the total quarterly revenue of nearly every single company on Earth. Hyperscalers, well that contributed 48.7 billion, up 102%. But AI, cloud, industrial, and enterprise, that contributed 40.3 billion, growing even faster at 138%. And this mix is important. The thesis is stronger when demand expands beyond four or five hyperscalers and reaches sovereign AI, neoclouds, enterprises, and industrial customers. We can see here the full income flowchart. It shows the same story. $89 billion from data center, another $7.2 billion from edge and very very impressive 72.1 billion of gross profit. This is not simply training demand. Inference reasoning models and agentic workloads consume compute every time a user asks a question or an agent completes a task. Turning deployed models into recurring infrastructure demand. And networking also matters because thousands of accelerators must behave like one computer. Nvidia's ability to sell the fabric around the GPU increases system content and makes the direct chip price comparison incomplete. The risk though is that data center now carries almost the entire investment case for Nvidia. Gaming can be healthy and automotive can grow, but neither can offset a serious pause in AI infrastructure. Now gaming grew strongly, professional visualization improved and automotive continued expanding. Their small beside data center, but they provide optionality rather than carrying the valuation. The near-term question is not whether AI demand exists, it's whether Nvidia can supply enough systems and preserve pricing power while competitors and customers build alternatives. And one post results analysis focus on the detail I care about the most. Sequential revenue still expanded around the high teens despite Nvidia already operating at an enormous scale. Yeah, look, I mean, when I uh see the quarterly sequential increase in revenue, they've had a trend of 20%, this quarter was 18%. So, you could say sequentially uh it didn't kind of uh stay on that 20% mark. But other than that, I mean, I think everything in the print suggests uh they're sold out. There is definitely a lot more demand than what they can supply. And I would be curious to look at the disclosure around hyperscaler exposure versus the non-hyperscaler piece which they typically disclose in their presentation and the earnings call >> and that distinction is vital. Nvidia is not clearing excess inventory with discounts is supply constrainted and management says demand exceeds what the company can currently deliver today. And the CPU opportunity it adds another layer. Management spoke about demand for roughly 20 billion CPU units and expects fiscal 28 CPU revenue to more than double. And Vera Rubin is not just a faster chip. It's an entire rack scale platform involving GPUs, CPUs, networking, memory, and software, increasing Nvidia's content per deployment. And when supply is the constraint, new architecture can lift both volume and average selling price. That is why revenue can keep compounding even after a near $100 billion quarter. And sold out also improves visibility, customers reserve capacity, and planned data centers years ahead, giving Nvidia a clearer order pipeline than an ordinary consumer semiconductor cycle would typically provide. Yet visibility is not the same as guaranteed cash. If customers delay facilities, struggle to finance projects or cannot monetize compute, orders can move right even when the strategic demand remains. This is why customer mix payment terms and the utilization of installed GPUs are as important as management saying every available system is spoken for. But supply constraints, they're not purely bullish. If customers cannot obtain enough hardware, revenue gets delayed. And if constrained components become more expensive, Nvidia may have to share the economics. And the biggest surprise, it came in fact on the earnings call. Nvidia projected around 70% sales growth for fiscal 28 and management explicitly described the outlook as supply constraints. In plain English, the forecast does not assume Nvidia can satisfy all available demand. The company's telling investors that the limiting factors capacity, not customer appetite, where the hyperscaler spending chart supports this view. Top five capital expenditure is estimated around $1.3 trillion in 27 with consensus originally expecting a plateau rather than another acceleration. And while Nvidia's guidance suggests that plateau may be too conservative if hyperscalers, neoclouds, and sovereign customers keep building 2028 it may be a continuation, not the peak. And this is the most valuation relevant part of the release. A one quarter beat adds cash. A longer growth runway changes the multiple than investors should apply to every future year. and consensus earning estimates now imply EPS growth of 94% for fiscal 27, 52% for 2028 and 33% for 2029 before slowing sharply. And at the current price, these estimates place Nvidia trading 23 time 27 earnings and only around 14.9 based on 2028. Now the hidden assumption is that earnings growth converts into cash if fiscal 28 revenue grows 70% but working capital absorbs a growing share. The low forward P will overstate the economic bargain. Conversely, if new capacity unlock shipments while margins hold, current estimates may still be too low. The asymmetry is what transform this report from a routine beat into a duration upgrade. This is why Nvidia can look expensive on trailing sales but inexpensive on forward earnings. The denominator is growing so quickly that conventional snapshots become stale almost immediately. And there was also a separate report that Nvidia agreed to acquire hugging face around $13 billion. At the time of these reports, neither company had publicly confirmed the deal. So I'm not going to build this into the valuation just yet, but strategically the logic is clear. Nvidia would move deeper into open-source models, developer tooling, and the software layer. The opportunity is to make CUDA and Nvidia infrastructure even more central to how developers build and distribute AI. The risk is paying a premium for an ecosystem that values independence. So for this episode, I'm going to treat the report acquisition as an optional upside and strategic context, not as a reason to raise the DCF. Now to the first real concern. Free cash flow was around $21 billion. It improved from a year earlier at 30 and a half, but it actually fell sharply from the previous quarter. And we can see management's own slide highlights both facts. Free cash increased year-over-year while shareholder capital returns reach a record $26 billion. Now Nvidia did say the sequential operating cash flow decline reflected higher working capital adjustments and cash taxes is not automatically structural but it deserves monitoring. Revenue cannot become cash if receivables grow indefinitely or customers require longer payment terms. For a DCF investor, cash conversion eventually matters more than adjusted EPS. And quarterly cash flow can be noisy because taxes, collections, inventory, and supplier payments do not line up perfectly with reported revenue. Is why avoid annualizing the $21 billion quarter mechanically. Instead, I compare cash conversion over several quarters with gross profit and net income. A timing issue should reverse. A structural financing problem. Thou persist and thou widen. And look the record capital return it also needs context. Repurching shares is useful only when the shares are below intrinsic value. Otherwise retaining cash or funding high return research that probably creates more value. And obviously one week quarter does not prove a trend. My warning level would be several quarters in which revenue accelerates but operating cash flow and free cash flow repeatedly lag. For now though profitability remains outstanding, the balance sheet is strong. The sensible response is to reduce confidence slightly, not to declare that the earnings quality are low. And the second risk is gross margin. Demand may be unlimited, but shareholders only benefit if Nvidia keeps a meaningful share of the value that it creates. >> Reading back over it, I'll just recount it. So, in the quarter gone, adjusted gross margins was 75% bang in line with expectations for the fiscal third, the guide is 74% plus or minus 50 basis points. I think consensus was 74.8%. 8% on the adjusted gross margin. Um, Kungjan just said that Nvidia has announced price increases. Maybe he knows something I don't, but all I'll do is say that Bloomberg had reported that the server makers were informing customers that starting in January 2027 that any system shipped from that point, both Blackwell, and Vera Rubin would be subject to higher prices of 15% or more. >> So, a 74% guide is not a collapse. But on $108 billion of quarterly revenue, one percentage point of gross margin, it represents more than $1 billion. And the long-term operating margin chart here is phenomenal. The dangers assuming that those margins can only rise while memory wafers, packaging, and entire rack systems become more complex and higher prices. They can protect profitability, but they also distribute revenue through TSM, memory suppliers, and server manufacturers. Not every dollar of system inflation belongs to Nvidia. There are three forces to watch. The price Nvidia charges, the cost of advanced memory and wafers, and the mix between individual components and complete systems. Complete rack systems can increase revenue per customer while carrying different pass through costs. So higher reported revenue does not automatically mean a higher percentage margin even if profit dollars rise. So when we come to look at valuation, I care more about total free cash flow than defending exactly 75%. A modest margin decline is acceptable if volume and cash profit keep compounding. My base case therefore should assume strong margins rather than eternally expanding margins. The DCF shouldn't require Nvidia to become more profitable every year forever. And the third risk is circular financing. Nvidia invests in AI companies and infrastructure providers. Those companies may then use capital to purchase Nvidia systems. Now, management rejected the criticism, arguing that Nvidia receives only $1 for roughly $8 of total AI infrastructure spending with most capital flowing elsewhere in the stack. And the defense is reasonable, but it does not make every investment risk-free. The real test is whether end customers generate enough revenue to support the infrastructure without continuous outside financing. And customer concentration that also remains high. Large hyperscalers can fund the spending, but their bargaining power and their internal chip programs create a different form of risk. But look, concentration cuts both ways. A small number of enormous customers can sign multi-year commitments and finance trillion dollar infrastructure plans. But one budgeting change can move billions of Nvidia revenue. The cleanest evidence against the circularity concern will be improving AI revenue at Nvidia's customers, cloud growth, software sales, agent usage, and productivity that the customers willingly pay for. And the fastest growing non-hypers scale segment helps broading from a few buyers to AI clouds, industrial firms, and enterprises. It makes the ecosystem less dependent on any single capital budget. My conclusion is neither ignore circular financing nor the revenue is fake. is a yellow flag requiring disclosure, cash collection, and enduser economics to keep improving. And look, AMD, Broadcom, Google, and major AI labs are all developing competitive or custom silicon. At Nvidia's scale, even modest share loss can affect the growth rate. But Nvidia's mode is not merely benchmark performance. It includes CUDA, networking, rack design, developer support, software libraries, and an installed base that reduces deployment risk. Customers may ultimately choose a cheaper chip for a narrow workload while still using Nvidia for frontier training, inference, and the most complex generalpurpose systems. Ultimately, custom accelerators do not need to beat Nvidia everywhere. They only need to be good enough and cheaper for a repeatable internal workload, which can limit Nvidia's pricing power at the margin. Nvidia's response is speed and integration, new architectures on a rapid cadence, more networking, more CPUs and software that makes the whole platform productive sooner. And that in turn means competition probably changes price mix and margins before it destroys demand. My valuation assumes gradual normalization, not permanent monopoly and not sudden obsolescence. The decisive evidence will be whether forward revenue and free cash flow growth remain far above the sector while gross margins hold near the low to mid70s. And look, Nvidia's current forward P sits around 21, well below its 5year average of 36. It makes the stock look inexpensive relative to itself. But relative to the semiconductor sector, however, Nvidia still trades at a large premium on sales, cash flow, and book value. Both statements can be true simultaneously. And if consent earnings arrives, the multiple falls to 14.9 on 2028 numbers and then down to 11.2 on 2029. That would be extremely attractive for a quality compounder. But these low multiples, they're not free. They assume EPS of $925 in fiscal 27, $149 in 2028. In other words, the stock is cheap if the growth estimates are right and expensive if the AI investment cycle peaks early. The valuation is a forecast disguised as a ratio. And we have analysts who have revised their price targets just this morning. JP Morgan raised its target to $320, citing stronger AI demand and the extending growth runway. Bank of America, well, they moved to $350, well above the pre-earnings market price. We have Mizuz, whose target was $315. That's closer to the broader Wall Street average with Goldman Sachs staying more cautious, $300 target and a neutral rating. Melius, they went even further, $420, demonstrating how dramatically a small change in long run assumptions can change the answer. And as I said earlier, the average price target from Wall Street sits at $313, lowerend $180, higher end 515. The enormous range here, that's the real message. Analyst targets, they can anchor expectations, but they don't tell us which cash flows, discount rate, or terminal growth they're using to justify the number. For that, we need the DCF. And before the valuation reveal, this portfolio manager framed the debate correctly. One earnings report cannot settle what Nvidia's margins and growth look like at the end of the decade. >> I don't actually think what the numbers are matters this time around. Um, and we're just going to have to listen to what he says. And I actually don't think he'll be able to say anything that's going to change bears or bull's mind. >> Why? Um because right now the argument isn't about what's happening this quarter or next quarter or even next year. It's what happens in 2029. What is the margin structure of this company? What is the growth rate of this company with with more competition? So that said, if you look at the whisper numbers on the street on 2028, this thing trades at a low double digit multiple and it is a coiled spring. >> She was right about the timing. This report strengthened the case for 2027 and 2028, the DCF must decide how much of the strength survives into 2029 and beyond. And because shareholders receive cash rather than adjusted earnings, I start with free cash flow, not a headline P multiple. Now, all three cases start up with their latest information, $21.4 billion diluted shares, just under a hundred billion of cash and around $ 38 billion of debt, where we use an 8% discount rate, 3% perpetual growth rate. The difference here though is going to be their near-term free cash flow. And after the explicit ramp I test here, as we can see the sensitivity table, low, medium, and high growth rates of 8, 12, and 16%. So the intrinsic values that we're going to get for the bare, bull, and base case, they're not predictions in the nearest dollar. They're boundaries that show what happens when the cash generation arrives slower or faster than expected. And 8% it is the optimistic part of the framework. A lower required return makes distant cash flows more valuable. So, I also recalculated the base case at 9 and 10%. Remember here though, the sensitivity is crucial. A DCF's not an oracle is a transparent way to expose what must be true for a target price to make sense. And let's jump in with the bare case first. Free cash flow rises to $158 billion in 27, 174 in 28, 191 in 29, and 211 billion in 2030. After that, the model uses the growth rate of 12% through to 2036, which gives us a price of $236 against the market value. While that's only around 6% upside after a strong earnings rally, while the margin of safety, it becomes very thin. We're talking around 6%. The reverse DCF here, it's coming in this bare setup at around 10.6% growth is achievable, but it leaves less room for execution errors, margin compression, or a higher discount rate. It's not a disaster scenario. is the scenario where Nvidia remains exceptional but cash flow normalizes faster than the most enthusiastic forecasts assume. In the base case, we have $180 billion in 2027, $27 in 28, then to 274 by 2030. And using the 12% growth rate, well, we get an intrinsic value of $34. And this in fact is the updated base case 304 from 295. And at today's price, it represents around 37% upside and a margin of safety sitting at 27%. The reverse DCF is coming in at only 4.8%. And we can see here at the low end, we get $255. At the higher, more aggressive end $362. Now, the discount rate warning at 9% using the exact same figure, while the mid rate actually drops to $248. And that implies an 11% margin of safety. Going one step further and increasing the discount rate to 10%, we get 209. So an overvaluation signal where we notice a premium around 6%. So the base case at 8% is only attractive if you accept both the cash flow runway and an 8% required return. A more conservative rate removes much of the upside. And then we get to the bull case where free cash flow reaches $200 billion in 2027 up to 300 billion by 2030. And we can see here again using the growth rate middle 12% we get an intrinsic value $332 and that does imply in fact quite a large 33% margin of safety with a reverse DCF incredibly small 2.8%. But remember the near-term cash flows do so much of the work and that is precisely why the explicit assumptions matter. And look if we were to use a 9% discount rate here we still see undervaluation. In fact we had a margin of safety of 18%. Going one step further, 10% on the most bullish case. It isn't too far off today's price. We can see two to 3% MOS. So, going back to the base case for Nvidia, we get $34. My verdict is bullish, not price insensitive. The quarter increased confidence in the duration of demand and justified raising my base rate to $34. The base case coming at $236 tells me not to chase blindly after a sharp move. A purchase near or below that level would offer a much more obvious MOS. So between $236 and $34 position size and required return matter above the base target. The investor increasingly depends on the bull case rather than merely owning a great business. And remember this heat map is from the pre-market is not the final verdict. Prices can change in the day. The cash flow assumptions that determines the long-term return. Now I want to know your number. Is $34 reasonable, too conservative, or too optimistic? And which assumptions would you change? If the breakdown helped you separate the company from the stock price, smash the like button, subscribe, hit the notification bell so you're aware when we drop future videos. And don't forget to sign up to our weekly newsletter. We drop one every single week covering severely undervalued stocks, what's going on in the market. You can click on the pin comment, read all of these straight away. More importantly, have a great day. I'll see you all on the next one.