KRW 100 million → about KRW 17.3 billion (17,311,235,946 KRW). The 17,211.24% cumulative return uses the September 22, 2026 close.
- Purchase → valuation
- →
- Initial amount
- 100,000,000 KRW
- NVDA purchase close
- $1.6255 split adjusted
- KRW 100 million after ten years
- 17,311,235,946 KRW
- Cumulative return
- 17,211.24%
- CAGR
- 67.44%
The NVIDIA result above is a KRW price return excluding dividends. The SPY and QQQ comparison below includes reinvested distributions. All results exclude personal taxes, trading and FX costs.
What would different investment amounts be worth?
Each row reruns the existing backtest engine at that initial KRW amount. Results include fractional-share rounding and leftover USD cash; amounts are rounded to the nearest KRW for display. On small screens, scroll the table sideways.
| Investment | Final value | Cumulative return | CAGR |
|---|---|---|---|
| 1,000,000 KRW | 173,112,359 KRW | 17,211.24% | 67.44% |
| 10,000,000 KRW | 1,731,123,595 KRW | 17,211.24% | 67.44% |
| 50,000,000 KRW | 8,655,617,973 KRW | 17,211.24% | 67.44% |
| 100,000,000 KRW | 17,311,235,946 KRW | 17,211.24% | 67.44% |
How the KRW 100 million investment changed over ten years
Last monthly observations trace the KRW value of the initial investment. Look for periods of growth and large reversals alongside the ending value above.
Read annual chart values as a table
| Date | KRW value |
|---|---|
| 2016-09-22 | 100,000,000 KRW |
| 2016-12-29 | 187,594,721 KRW |
| 2017-12-28 | 294,573,203 KRW |
| 2018-12-28 | 207,862,340 KRW |
| 2019-12-30 | 374,501,963 KRW |
| 2020-12-30 | 796,259,371 KRW |
| 2021-12-30 | 1,961,166,460 KRW |
| 2022-12-29 | 1,029,627,958 KRW |
| 2023-12-28 | 3,556,587,083 KRW |
| 2024-12-30 | 11,287,215,895 KRW |
| 2025-12-30 | 15,053,066,800 KRW |
| 2026-09-22 | 17,311,235,946 KRW |
Why holding on for ten years was difficult
Today we know that the initial KRW 100 million became about KRW 17.3 billion. At each point along the way, the next rise, the depth of the next decline and the timing of recovery were still unknown. To think about holding NVIDIA for ten years, consider the decisions an investor faced before knowing that ending value.
The feeling that you had already earned enough
Imagine an investment that has doubled or tripled. Selling would turn a gain on the screen into money you could use, perhaps for a home or another goal. Even while the business keeps growing, the thought of losing a gain already made can make taking profit appealing.
A large gain can also leave one stock accounting for more of your assets than you originally intended. The question then becomes how much of your wealth you want tied to NVIDIA. Keeping the entire holding requires revisiting both the business case and the weight of that investment.
Watching gains disappear from the peak
In the stored daily closes, the USD share price fell 66.36% from its peak to the trough, leaving it at roughly one-third of the peak price. The stored series first exceeded that peak again on . This decline is measured in the USD share price; the chart above follows the KRW investment value including exchange-rate changes.
Even if the price remains above your purchase price, money lost from the peak can feel like a loss. You might wish you had sold earlier and worry that the next fall will erase the remaining gain. A backtest shows the recovery date; a holder living through the decline still had to decide with that date unknown.
Questioning the business when results weakened
There were business changes behind the uncertainty too. In the FY2023 third-quarter results released on November 16, 2022, gaming revenue fell 51% year over year while data-center revenue grew 31%. Weakness in an established business and signs of new growth existed within the same company.
An investor had to consider how long gaming weakness might last and whether data-center growth could support the wider business. Following customer demand, products and earnings would help them revisit their reason for owning the shares. Before knowing the eventual return, could you have explained why you still wanted to own NVIDIA through its business and results?
Why did investors buy NVIDIA ten years ago?
The investment case in September 2016 centered on gaming and data-center growth. Here is what the information public by September 22 showed.
- An existing gaming GPU business. NVIDIA’s August 11, 2016 FY2017 Q2 release described the Pascal GeForce GTX 1080, 1070 and 1060 launches and total quarterly revenue of $1.428 billion, up 24% year on year. High-performance graphics was already a paying market.
- CUDA and deep-learning software. NVIDIA’s April 2016 announcement presented Tesla P100 alongside CUDA 8 developer tools. These supported a software-backed GPU opportunity without specifying its eventual size.
- A data center opportunity. The August 2016 release presented Tesla P100 for data centers and a program supporting deep-learning startups. The products and use cases available at the time showed GPUs expanding beyond gaming.
- An automotive option. NVIDIA’s February 2016 release presented DRIVE PX 2 and an autonomous-driving pilot. A pilot was not evidence of large-scale autonomous-vehicle sales.
Why might an investor have passed in 2016?
- Price expectations. Contemporaneous reporting noted the stock had already climbed sharply. Growth expectations may already have been priced in; a promising business did not necessarily mean an attractive entry price.
- Competition and manufacturing. NVIDIA’s 2016 release itself listed technological change, competition and reliance on outside manufacturers among its risks.
- Uncertain staying power. The same release warned that demand for new products and market adoption could differ from expectations. Investors had reason to study gaming results alongside data center demand.
Did the 2016 investment case play out?
NVIDIA’s February 25, 2026 FY2026 release reported full-year data center revenue of $193.7 billion, gaming revenue of $16.0 billion and automotive revenue of $2.3 billion. The direction of GPU-accelerated data center growth proved right, and gaming grew too. The scale and pace of later AI demand were not knowable from the 2016 numbers. Automotive grew but did not become a comparably large revenue driver; its segment is not autonomous-driving software revenue alone.
How did SPY and QQQ compare?
SPY tracks the S&P 500 and QQQ tracks the Nasdaq-100. This comparison uses the same KRW 100 million and the same purchase and valuation dates. Scroll the table sideways on small screens.
NVDA: share price + FX, dividends excluded. SPY and QQQ: distribution-reinvested adjusted close + FX, before taxes and costs. Same KRW 100 million and endpoint dates.
| Asset | Dividend basis | Final value | Cumulative return | CAGR |
|---|---|---|---|---|
| NVIDIA | Excluded | 17,311,235,946 KRW | 17,211.24% | 67.44% |
| S&P 500 ETF (SPY) | Reinvested distributions | 511,753,907 KRW | 411.75% | 17.74% |
| QQQ | Reinvested distributions | 826,080,173 KRW | 726.08% | 23.51% |
SPY and QQQ use the start-to-end ratio of stored adjusted closes, which incorporates pretax distribution reinvestment. It does not reproduce account-level payment dates, withholding or trading costs. NVIDIA’s hindsight result does not erase single-stock concentration risk. ETF rows in the ten-year comparison hub use a separate price-only basis.
Comparison basis
The KRW results cover September 22, 2016 to September 22, 2026 without further contributions: NVDA dividends excluded; comparison ETF distributions reinvested. Personal taxes, trading and FX costs, and inflation are excluded. Prices and same-record-date USD/KRW observations are combined.
Calculation details and source records
Prices, splits, FX and calculation basis
The September 22, 2016 actual NVDA close of $65.02 becomes $1.6255 after dividing by the subsequent 2021 four-for-one and 2024 ten-for-one splits. The stored September 22, 2026 close is $228.87. Splits adjust shares and per-share prices; they create no return by themselves.
The KRW principal converts to USD at the stored purchase-date 1103.3 KRW/USD, and the ending USD holding converts back at 1356.5 KRW/USD. The NVDA body and chart exclude dividends; the SPY and QQQ comparison rows incorporate reinvested distributions. Additional contributions, personal taxes, trading and FX costs, and inflation are excluded. CAGR annualizes the full-period return; it does not imply equal returns every year.
Detailed calculation rules
For the NVIDIA result, the existing backtest engine floors US share quantity to eight decimal places and retains leftover USD cash. Valuations use two decimals and returns six decimals. The ETF comparison instead uses the adjusted-close ratio with the same endpoint FX rates. Display rounds KRW to whole won and percentages to two decimals with HALF_UP. CAGR uses actual elapsed days divided by 365.2425.
Calculation notes and limitations
- NVIDIA uses stored split-adjusted closes; SPY and QQQ use dividend- and split-adjusted closes. FX uses same-record-date USD/KRW observations. These are not live prices, executable FX quotes or verified official fixings, and US closes and Korean FX rates are not simultaneous quotes.
- There is an observation gap from to . Full trading-calendar reconciliation is unfinished, so the MDD above is the largest observed USD share-price decline. Missing FX observations also prevent a daily KRW MDD.
- The September 2026 NVDA close was checked against Nasdaq history. The 2016 purchase close and every daily exchange record have not been independently reconciled. Dividends, taxes, costs and inflation affect actual results.
Reference links: Nasdaq NVDA history · SPY issuer · QQQ issuer. The 2016 investment case, split announcements and later results link to NVIDIA releases beside each claim.
Frequently asked questions
What was NVIDIA's ten-year return?
For the fixed 2016-09-22 to 2026-09-22 interval, KRW 100 million became 17,311,235,946 KRW. Cumulative return was 17,211.24% and CAGR was 67.44%. This is a KRW model excluding dividends, investor taxes and trading costs.
Why is the 2016 NVIDIA purchase price shown as $1.6255?
The actual close was $65.02. Dividing by the subsequent 2021 four-for-one and 2024 ten-for-one splits gives a comparable $1.6255 price. A split alone did not create a gain.
Do the KRW results include currency and dividends?
They use stored rates of KRW 1,103.3/USD at purchase and KRW 1,356.5/USD at valuation. The NVDA result excludes dividends; the SPY and QQQ comparison rows use adjusted closes that incorporate distribution reinvestment. Executable FX spreads, taxes and fees are excluded.
Is the 66.36% drawdown a KRW account loss?
No. It is the largest observed decline across 2508 stored split-adjusted NVDA USD closes. It is not a KRW portfolio MDD or a certified full-period maximum because some price and FX records are missing.
What should I examine when backtesting NVIDIA?
Compare NVDA and SPY across starting dates and examine returns, maximum drawdown and recovery. Use the results alongside study of the data-center business to consider a holding period and allocation.
Owning a stake in NVIDIA’s growth
I believe we should invest some of what we earn. Owning shares lets our money participate as companies develop technologies and markets. NVIDIA illustrates the meaning of business ownership. Study the business when choosing a stock, and keep examining customer demand and earnings while holding it. Backtests help you explore the declines you would have lived through alongside the final return.
Try a different period or amount
Compare NVIDIA with SPY across several starting dates. Look at returns, drawdowns and recovery times to plan how you would hold the stock.
Open NVIDIA backtestThis article presents the author’s investment opinion and historical observations. Returns and losses depend on the asset and investment period.