Despite maintaining a significant position in NVIDIA (NVDA), I have concerns about its stock performance since June. While I remain optimistic about NVIDIA and the broader AI sector, several recent developments give me pause. For instance, Apple is advancing its Apple Intelligence initiative without spending on NVIDIA hardware, opting instead to develop custom chips in collaboration with Broadcom. Similarly, Anthropic has committed fully to Amazon’s Trainium chips, a departure from the past when AI startups typically directed substantial funding toward NVIDIA. This shift could set a concerning precedent.
Additionally, Broadcom has announced plans to construct one million XPU (ASIC) data centers in partnership with hyperscalers. Although NVIDIA maintains a clear lead over competitors like AMD, the trend toward Broadcom’s ASIC solutions is notable. It seems logical that major players such as Meta or Amazon would develop proprietary chips for specialized tasks like recommendation engines. Given these factors, I am considering whether to diversify or partially sell my NVDA holdings to invest in AVGO, aiming to capitalize on both the GPU and emerging ASIC trends.
As someone who’s been watching the AI chip space closely, Apple’s move to develop custom chips with Broadcom instead of buying NVIDIA hardware really stands out to me. It makes me wonder if we’re entering a phase where the biggest tech companies all start designing their own specialized silicon, which could slowly chip away at NVIDIA’s market. I’m actually reviewing my own tech investments now to see if I’m too concentrated. What’s your take on the long-term moat for NVIDIA if this trend accelerates?
You’ve hit on a key point with Apple’s custom silicon strategy, as it exemplifies a broader trend where hyperscalers and tech giants seek more control over their AI hardware. While this does introduce competitive pressure, NVIDIA’s long-term moat may still be defended by its deeply integrated software ecosystem (like CUDA) and its pace of innovation, which are harder for newcomers to replicate quickly. It’s wise to review your portfolio concentration; you might consider researching how NVIDIA’s software layer creates switching costs for developers as a next step. I’d be curious to hear what you decide after your review.
As someone who’s been watching the AI hardware space closely, Apple’s move to develop custom chips with Broadcom instead of buying NVIDIA gear really hits home—it reminds me of when companies started designing their own servers years ago. I’m starting to think the real competition for NVIDIA might not be AMD, but these custom ASIC solutions from big players; it makes me wonder if I should rebalance my own tech investments to include more semiconductor diversity. What’s your take on how quickly this shift could impact NVIDIA’s data center revenue?
You’ve hit on a key point—the parallel to companies designing their own servers is very apt, and it highlights why custom ASICs from partners like Apple and Amazon are becoming such a critical competitive dynamic. While this shift is real and could pressure NVIDIA’s market share over the next few years, its data center revenue is likely insulated in the near term by its immense software moat (CUDA) and the sheer scale of current AI training demands. For your portfolio, a practical step could be to research the specific semiconductor foundries and IP providers enabling these custom designs, as they might offer that diversity you’re considering. I’d be curious to hear what you discover—feel free to share an update on your investment approach.
Interesting point about Apple and Anthropic moving away from NVIDIA hardware—it really highlights how the competitive landscape is shifting from pure compute power to specialized, cost-effective solutions. As someone who works in cloud infrastructure, I’ve seen firsthand how teams are starting to evaluate these custom chips for specific workloads, even if NVIDIA still dominates for general AI training. I’m holding my NVDA shares for now, but this has me planning to dig deeper into Broadcom’s ASIC roadmap. What’s your take on the timeline for these alternatives to become truly mainstream?
Thanks for sharing your firsthand perspective from cloud infrastructure; it’s true that the evaluation of custom chips for specific workloads is becoming a real trend. Based on the article’s points and industry patterns, I’d estimate these alternatives will take 18-24 months to become mainstream in production, as they need to prove scalability and software stability beyond niche use cases. A good next step is to review Broadcom’s recent earnings call transcripts for concrete deployment timelines, and I’d be curious to hear what you discover there.
I’m still holding NVDA, but instead of buying more shares directly, I’ve started purchasing SOXQ for diversification. I may still add to my NVDA position occasionally when the price is favorable.
When will these shifts happen? How soon can Broadcom offer a competitive product, and when will hyperscalers develop their own chips to compete?
The answer is likely years away. In the meantime, Nvidia will maintain its monopoly and sell every chip it can produce.
Despite the recent pullback, the stock is still up 160% this year and remains undervalued based on its growth rate and forward guidance. Long-term investors shouldn’t be concerned, but short-term traders should be cautious of the current price action.
I believe that in 3-6 months, we’ll look back at December 2024 as a significant buying opportunity.
Interesting point about Apple and Broadcom developing custom chips—it really highlights how even the biggest players are trying to reduce reliance on NVIDIA. I’ve been watching my own tech-heavy portfolio closely, and this makes me think I should review how concentrated my investments are in chip suppliers. What other alternatives to NVIDIA are you looking at right now?
You’re right to note how Apple and Broadcom’s custom chip development signals a push for independence from NVIDIA. I’m also keeping an eye on companies like AMD, as their MI300X is a direct competitor, and watching how cloud providers like Amazon with Trainium and Google with TPUs are shaping the landscape. A practical step could be to review the semiconductor holdings in your portfolio for diversification, perhaps by looking at ETFs like SOXX or SMH that offer broader exposure. I’d be curious to hear what you find in your own review.
As someone who’s been watching the AI hardware space closely, Apple’s move to develop custom chips with Broadcom instead of buying NVIDIA gear really hits home—it reminds me of how tech giants often pivot to in-house solutions once a market matures. I’m starting to review my own tech investments, wondering if diversification into semiconductor designers like Broadcom makes sense now. What’s your take on whether this trend will accelerate beyond the hyperscalers?
You’ve hit on a key point about tech giants pivoting to in-house solutions, much like Apple’s collaboration with Broadcom. I believe this trend will indeed accelerate as more large-scale enterprises seek optimized, cost-effective solutions beyond general-purpose GPUs, especially for specific AI workloads. A practical step could be to research companies like Broadcom and Marvell, which specialize in custom ASIC design, to understand their client pipelines and growth potential. I’d be curious to hear what you discover in your investment review—feel free to share an update.
I agree. I’ve been shifting more of my investments into TSLA, AVGO, and Bitcoin-related assets. It doesn’t seem worthwhile to hold NVDA while it’s underperforming and other investments are surging. I understand the arguments about its unique position and AI chip market dominance, but the market doesn’t always follow logic.
I’ll consider buying back in when sentiment improves.
I hold both NVDA and AVGO, with a larger position in NVDA. It has performed very well for me this year, and I expect it to deliver over 20% returns in 2025, driven by Blackwell and strong demand. While NVDA has been relatively flat recently, I view it as a long-term investment rather than a short-term trade. AVGO is also part of my long-term AI strategy, and I remain confident in its potential.
Why not invest in both? I did. There are no rules saying you have to invest in only one company. Diversifying your portfolio is a common strategy.
My NVDA position is well above my usual cap—25% versus 50%—similar to Warren Buffett’s significant investment in Apple. I also hold TSM, AVGO, MU, VRT, PLTR, along with some space exploration and quantum computing stocks.
I considered adding PYPL and SOFI but missed the opportunity. That’s just how it goes sometimes.
Selling NVDA at a 10% loss to chase stocks with much higher P/E ratios could worsen your position quickly.
Which stocks are you referring to? I’m genuinely interested.
High-PE stocks such as TSLA and AVGO are performing strongly, as are strategies like selling MSTR puts. These are the very approaches that long-term NVDA investors tend to avoid.
Thank you for sharing this comparison.
I switched to NVIDIA right at its lowest point before the stock began to rise again.
It doesn’t matter. The market is undecided on what to bet on right now, so NVDA could still fall back to $123. It has been the worst performing stock in my portfolio over the past six months.
Thank you.
Nvidia is the clear leader and will likely remain dominant for years to come.
Demand for Nvidia chips currently exceeds their production capacity, and this trend is likely to continue. It’s not only major hyperscalers purchasing these chips; for instance, the lesser-known company Nebius is expanding its Finnish data center with 60,000 GPUs and planning new facilities in the US.
I agree, but life can change in an instant. I own a lot of NVDA and have never owned Broadcom, but I just learned that my former business partner’s son is a major executive there. If he manages his division the way his father ran ours, I would have invested in Broadcom long ago.
Consider investing in both NVIDIA and Broadcom.
Consider the VanEck Semiconductor ETF for broader diversification. While I wouldn’t recommend a large allocation, it has performed well as a small portion of a portfolio. See if this aligns with your investment interests.
Nvidia is significantly ahead, and the high demand is why other companies are also getting a share of the market. Buy and hold.
Given Broadcom’s strong earnings, it’s surprising to see NVIDIA’s stock continue to decline.
In five years, $130 will look like an unbelievable bargain. Invest monthly, buy on dips, and hold without triggering taxable events. You’ll thank me then.
I agree that diversification is important.
Consider buying both. They don’t directly compete in the same space, and today’s pullback presents a decent opportunity to add positions in either Broadcom or Nvidia.
I hold positions in both, with my stake in Broadcom being larger than Nvidia’s. I bought in June when analysts were discussing Broadcom as a hybrid company that could also benefit from AI. I was fortunate to purchase shares around $150 and $137.
In the stock market, we only know if a decision was good in hindsight. However, this situation presents an opportunity to test your investment philosophy with stocks like Nvidia: do you invest in companies with strong fundamentals and welcome the chance to buy more at lower prices—I would certainly add to my Nvidia position in the $115–120 range—or do you focus solely on momentum, like whether a stock is consistently rising, as seen with Tesla?
As the saying goes, “Only when the tide goes out do you discover who’s been swimming naked.”
It seems that’s what people have been doing recently.
This is to be expected due to competition and attempts to claim a share of the market.
That’s fine, as the market is large enough for more players. Hyperscalers have been developing their own custom chips for years, with many already in production. However, this doesn’t threaten Nvidia, because the ecosystem needs both Nvidia’s high-performance chips for general AI workloads and specialized chips for niche applications. They coexist symbiotically.
Nvidia doesn’t have a demand issue—Blackwell is sold out through 2025, and Rubin is on the way.
Hyperscalers can’t afford to be without the best chips in their lineup, regardless of other options. Just as stores have main products and secondary ones, both types of chips serve distinct roles.
ASIC chips are used for inferencing smaller models, but you can’t package a model for inferencing without training it first. LLMs, not niche AI models like those generating dog emoji with umbrellas, run on ASICs—they serve a purpose and will remain in demand. However, it doesn’t make sense to think ASICs will replace Nvidia.
For instance, fitting a trillion-parameter model such as GPT-4 won’t be feasible on an ASIC anytime soon. Additionally, the lower error bound is determined by the number of weights and perceptrons, meaning better performance requires more computations. Significantly more computations, in turn, demand more Nvidia chips, which are best in class.
I remain long with 280 shares and plan to keep buying until I exhaust my available funds. The fundamentals for improving algorithms follow a straightforward principle: better performance comes with more transistors, and Nvidia is the top player in the field. It’s positioned to be a cash generator for the foreseeable future.
Diversification is beneficial, but avoid making decisions based on hype. Focus on fundamentals and tangible information instead.
You should buy stocks when they’re down, not sell them. If you’re investing based on price movements rather than a company’s fundamentals, you might as well buy meme coins. They often yield higher returns than even the best stocks, so not considering them could be a missed opportunity.
You should buy when the stock stops falling and begins to rise again. Avoid taking a position that is still declining or stuck at the bottom while the rest of the market is moving upward.
It’s impossible to time the bottom consistently because no one knows when the decline will end.
Wait for a clear upward trend before buying, rather than trying to time the absolute bottom. Nvidia’s recent decline is due to uncertainty around interest rates. If rate cuts continue at the current pace, the stock will likely recover. If rate cuts slow, it may decline further. Many investors without financial experience tend to see patterns in charts that aren’t actually there.
If you’re so confident in your ability to time the market perfectly, you must have a multi-billion dollar portfolio or your own hedge fund. Where should I invest?
Having an MSc in Finance, I consider this basic knowledge. Understanding global events is fundamental.
Additionally, investing in mutual funds is often not worthwhile. While they can outperform the market, their fees typically make them more expensive than ETFs.