Two chipmakers, opposite decades
NVDA vs INTC
In January 2015 these were both large, profitable, widely held semiconductor companies, and Intel was the safer of the two by almost any measure an investor would have used. A contribution plan started then produced results that differ by more than a factor of thirty.
#What this plan actually returned
These are real results, not estimates. We ran $10,000 up front plus $500 a month from January 1, 2015 to September 9, 2026 (11.7 years), with dividends reinvested, against actual daily closing prices for each fund.
| Ticker | Invested | Final value | Total return | Ann. return | Worst drop |
|---|---|---|---|---|---|
| NVDA | $80,000 | $9,461,348 | +11726.68% | +68.53% | -66.34% |
| INTC | $80,000 | $253,162 | +216.45% | +16.52% | -70.80% |
Ann. return is the internal rate of return (IRR), which accounts for how long each individual contribution was invested, so it is lower than the headline total return on a plan that kept adding money. Worst drop is the deepest peak-to-trough decline in the fund itself during this window, measured on daily closes.
Figures current as of . Results change as markets move.
Pre-tax and excluding brokerage commissions. See How It Works for the full methodology, or re-run this exact comparison with your own numbers.
#Start with what the numbers do not say
The obvious reading is that Intel was a disaster. It was not. INTC turned $80,000 of contributions into roughly $252,000, an annualized return of about 16 percent. In isolation that is a good outcome, comfortably ahead of what a broad-market fund returned over a similar period.
The comparison is brutal only because the other column exists. NVDA produced roughly $9.5 million on the same contributions, an annualized return near 69 percent. Both stocks made money. One of them made so much more that the other looks like a failure by proximity.
#Why Intel was the reasonable choice in 2015
It is worth reconstructing the decision honestly, because hindsight makes it look obvious and it was not. In early 2015 Intel dominated CPUs for personal computers and servers, owned its own leading-edge manufacturing, and paid a substantial dividend. It was the larger company and the more established one.
NVIDIA at that point primarily sold graphics cards to gamers. Its use in machine learning was an emerging niche, not a thesis most investors held. Buying NVDA over INTC in 2015 required either an unusually early read on parallel computing or a willingness to hold a more volatile, less established company for reasons that would not have sounded rigorous at the time.
What actually happened over the following decade was a set of specific developments: Intel repeatedly delayed its process node transitions and lost its manufacturing lead, while NVIDIA’s parallel architecture turned out to be the substrate for deep learning and then for generative AI. Neither outcome was knowable from a stock screener in 2015.
#The drawdown column tells the real story
Both stocks fell roughly two thirds at their worst, but the shape of those declines could not be more different, and the dates make it obvious.
- NVDA fell about 66 percent from November 2021 to October 2022, then recovered and went on to substantial new highs. It was a violent interruption inside a continuing advance.
- INTC fell about 71 percent from April 2021 to April 2025. That is a four-year decline, not a crash. There was no single moment of panic to hold through, just a long sequence of quarters in which the company lost ground.
A drawdown that resolves in eleven months and a drawdown that grinds on for four years demand completely different things from an investor. The first tests your nerve. The second tests whether your original reasoning was correct, over and over, while the evidence accumulates against it. Most people find the second far harder, and a single “maximum drawdown” percentage treats them as equivalent.
#What dollar-cost averaging did in each case
Steady contributions behave very differently against these two paths. Into NVDA, every contribution made during the 2022 decline bought shares that participated fully in the subsequent recovery, which amplified the final result well beyond what a single lump sum at the start would have produced.
Into Intel, contributions made throughout the 2021 to 2025 decline kept buying into a position that continued to fall. Dollar-cost averaging lowers your average cost, but it does not distinguish between a temporary dislocation and a durable deterioration. Buying more of something on the way down helps enormously in the first case and compounds the problem in the second, and you cannot tell which one you are in until afterward.
This is the honest limitation of the strategy, and it applies specifically to individual companies. An index has no equivalent failure mode, because it replaces declining constituents mechanically rather than holding them to zero.
#What to test
- Run this window ending in early 2021, before Intel's decline began. The ranking is far closer, which shows how much the end date determines the story.
- Add VOO as a third ticker so both stocks are measured against simply owning the market.
- Add a few other 2015-era semiconductor names to see the full range of outcomes rather than the two extremes.
- Compare the drawdown durations, not just the depths. The dates are in the table above.
Change the contribution amount, schedule, or date range and re-run this comparison against the same historical data.
Open this in the calculator →