Why Nvidia is a "Disaster" *Despite* Beating Expectations

ShenGuang
08-30

Leading AI-relevant chipmaker Nvidia Inc's (ticker: $NVIDIA Corp(NVDA)$) earnings for its second quarter (Q2) beat analysts' consensus estimates by delivering $30.04 billion versus an expectation of $28.7 billion and adjusted earnings per share (EPS) of $0.68 versus an expectation of $0.64. Despite projecting $32.5 billion in revenue over the current quarter versus expectations of $31.7 billion and Chief Financial Officer Colette Kress expressing confidence in the company being able to several billion dollars in value of its next-generation Blackwell products (currently being tested as samples) in its fourth quarter, the company's stock dropped by 8% in extended trading. When markets opened, the stock fell another 6.38%.

Trend Drilldown

As of the first half (H1) of its ongoing fiscal year (FY) 2025, trends indicate that the company's explosive growth in FY 2024 isn't going to be repeated in the current FY:

Note: Net Income Per Share has been restated with the stock split factored in.

Outside of CFO Kress' projections on sales of Blackwell products, the company is looking to close overall revenues at a 16% deficit relative to the previous year albeit with net income per share effectively closing the year with a 12% over the previous FY. 

The previous FY was the beginning (and possibly the end) of the "AI Stock Boom" which Nvidia led. A 126% rise in revenue was accompanied by an over fourfold boom in net income per share while operating expenses more-or-less ran flat as expenses incurred in research and sales preparation in the previous few years paid off. While the stock's high valuation does create a massive bump in stock-based compensation, it clocks in at 4% of net revenue - far lower than the levels seen over the past six FYs. 

Current-generation hardware inventory is over twice that of the previous FY, which could possibly indicate that either demand is flagging or that expected order volume might have been an overshooting on actual demand. 

As the company went "corporate", it is increasingly dependent on a small number of clients: Microsoft ( $Microsoft(MSFT)$), Meta Platforms ($Meta Platforms, Inc.(META)$), Alphabet and Amazon together contribute over 40% of Nvidia's total revenue. 

Factors for Consideration

Currently, the breadth of investing public is pondering as to whether the speculation-level high valuations imparting on certain names in the semiconductor space can be deemed sustainable. As the article about Nvidia's top fabrication partner TSMC indicated, institutional investors have grown skeptical over whether promises of high capital expenditure by "Big Tech" are justifiable given marginal contributions by AI-driven processes in supplanting human labour. This particularly affects Nvidia, given its high exposure to a handful of clients leading the consumption of its products. 

The company's centering of focus on Blackwell chips also has a significant market limitation factor: as it stands, the current-generation A100, H100, A800 and H800 chips are forbidden from being exported to China, which is a key market for the company - standing at almost 17% by the end of FY 2024. In a similar vein, the company's Blackwell chips are almost certain to be banned as well, given their enhanced AI-relevant processing capability. While China's tech giants have doubled capital spending on AI infrastructure in the year so far, the company hobbled in terms of client breadth and ever-more dependent on its existing small band of clients.

Roughly about 10% of the company's Q2 revenue ($3.7 billion) came from the sale of networking products. In this space, bête noire and (almost literal) cousin AMD, Inc ($Advanced Micro Devices(AMD)$) is aiming to close the gap: roughly ten days before Nvidia's earnings release, AMD announced a $4.9 billion acquisition of ZT Systems, a leading AI compute infrastructure and storage solutions provider in a bid to provide enhanced services to datacenter clients. As the article on AMD's earnings indicated, the company's footprint among corporate clients had substantially increased this current FY. 

With a host of startups (such as Sambacore) scrambling to provide next-generation computing hardware to select set of clients outside of China and existing solutions in the market largely being deemed more than sufficient to handle current AI-relevant processing loads and a dimming outlook on a repeat of explosive capital investments, the company's ability to have sustained sales volumes over a long horizon comes into question. 

In Conclusion

Without sustained sales volumes in the outlook, conviction stills and high valuations are likely to be tested. While "instituitionals" aren't likely to start offloading in bulk any time soon, there is very little incentive to acquire more.  However, the likelihood of a paring of exposure in a "risk-off" exercise in favour of a more secular exposure to the universe of investable assets increases. 

Despite splurging $15.4 billion in H1 2025, $7.5 billion of board-authorized capital for share buybacks had remained unutilized in Q2 2025. The company is topping this amount up with an additional $50 billion for share buybacks to prop up valuations. While this might indeed help ameliorate the speed with which the share value will depreciate, given the loss of sky-high conviction, this doesn't alter the forward outlook: as far as the company's products are concerned, it's a buyer's market with a small number of buyers and an almost-equal (if not greater) number of sellers. 

Buyer-seller dynamics are very much a play in motion with no great incentive to either "buy the dip" or "sell the crest". At the same time, there is precious little reason to load up and hold over a long term. While the company did start offering dividends as of FY 2023, the dividend yield is currently vanishingly small and of little incentive for dividend-focused investors.  

While the company remains an integral part of the hardware industry with great products and services, it is entirely divorced from the stock's valuation which remains massively distorted on account of the massive pile-on across much of FY 2023 and FY 2024. 

***********

Note: Investors with access to European exchanges can consider Exchange-Traded Products to boost gains on short-term stock trajectories. The “+3x Long Nvidia ETP” (LSE ticker: NVD3) gives a 3X daily-rebalanced exposure to the upside of the stock while the “-3x Short Nvidia ETP” (LSE ticker: NV3S) does the same for the downside.

PS: The 3x ETP has just started falling on account of the stock’s recent downfall:

while the -3X ETP has recently spiked.

***********

For broader articles that deep-dives into business and culture in Asia, visit asianomics.substack.com. The latest article discusses the spate of Mercedes Benz BEVs spontaneously combusting across Asia and trends in Benz’ EV sales. Also recently published is the rationale behind my commentary featured in Bloomberg, Fortune Magazine and CoinTelegraph.

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Comments

  • Dr Rck
    08-30
    Dr Rck
    Will normal investors be able to understand this to know it not worth to invest in right after the accountant ? Of course not, obviously rigged since too much money is at hand to buy at the face of higher than expected metrics reported all around? Rigged, no?
    • ShenGuang
      A lot of people bought heavily into this generally-stable high-quality company as the AI craze began. For the stock to keep high, people have to *continuously* buy into the AI story, which doesn't seem to be happening.
  • Stonkedaddy
    08-31
    Stonkedaddy
    A buyers market… interesting take given the critical nature of being first on new technology here and Nvidia technological moat. You don’t get over 50% margins in a “buyerss” market.
    • ShenGuang
      Depends. Datacenters are discovering that mixing in cheaper GPUs (such as AMD) optimize costs without significant impact on total performance. So there's a ceiling on demand while upcoming rival products will make product pricing more competitive.
  • JessieTheresa
    08-30
    JessieTheresa
    It's surprising to see Nvidia's stock drop despite beating expectations.
    • ShenGuang
      Yeah, the drop was sudden. I expected a "cool-off" shortly afterwards but the warnings were coming in from Wall Street since June (as shown in the TSMC article)
Leave a comment
6
12