Everyone Thinks They Know What a Stock Split Does to NVIDIA. Almost Nobody Does.

Zero. That’s the dollar amount of fundamental value NVIDIA’s (NVDA) 10-for-1 stock split created on June 10, 2024. Not approximately zero. Exactly zero.

I see the same mistakes over and over. They’re worth naming precisely.

I’ll confess something first: early in my career at Morgan Stanley, I spent real desk time modeling post-split price trajectories as if the split itself were a catalyst. I was wrong. It took me years — and watching retail investors repeat my mistake across six NVIDIA splits since 2000, amounting to a cumulative 480-for-1 adjustment — to understand that the cost of a stock split isn’t financial. It’s cognitive.

The split eats your attention, and attention is the scarcest resource an investor has.

Yahoo Finance reported NVIDIA’s pre-split Friday close at $1,208.88, which became $120.88 on a split-adjusted basis the following Monday. The 2024 split changed only the share count and nominal share price — it didn’t change NVIDIA’s market value or fundamentals. Not revenue. Not margins. Not competitive position.

If you owned one share worth $1,208.88 before June 10, 2024, you owned ten shares worth $120.88 each after. Same $1,208.88. NVIDIA took a $100 bill and handed you ten $10 bills.

The crowd applauded — reasonably, if you believe splits signal management confidence and open shares to a wider buyer base. That belief isn’t irrational. It’s just incomplete. The applause still cost something — I’ll get to what.

Buried in the same May 22, 2024, press release — buried is the operative word — sat an $80 billion share buyback authorization. This is the number that should have stopped every investor cold.

A buyback of that magnitude means NVIDIA’s board authorized systematic retirement of its own shares from the open market. Fewer shares outstanding means each remaining share represents a larger slice of the same earnings base. EPS rises mechanically. NVIDIA later expanded capital returns further, raising the dividend to $0.25 per share from $0.01 and authorizing an additional $80 billion buyback, with $39 billion remaining from a prior authorization.

Two genuine, quantifiable signals sat in the same document as the cosmetic split. They received a fraction of the coverage. That’s not primarily a journalism failure — the incentive to cover the visually dramatic event is obvious. It’s a cognition failure on the receiving end.

The visual drama of a price moving from $1,208 to $120 is immediate and visceral; modeling EPS accretion across an $80 billion buyback requires arithmetic and patience — and patience doesn’t generate clicks.

I hear it constantly: “But splits make shares more accessible to retail investors, and that creates demand.” As FINRA’s investor education materials explain plainly, a stock split adjusts share count and price proportionally. Period. More shares at a lower nominal price can improve trading liquidity and narrow bid-ask spreads. That’s a convenience. It isn’t value.

NVIDIA’s prior split was a 4-for-1 in July 2021. The stock ran hard afterward — but so did every semiconductor name in 2021, because the capex cycle was accelerating and supply constraints were tightening. Attributing post-split gains to the split itself is the same error as crediting rain to your car wash.

This distinction matters most for investors who bought post-split and held. Yes, they captured real gains if NVIDIA’s business performance continued. But those gains came from Jensen Huang’s execution on data center demand, not from the share count doubling or decupling.

Here’s the truth: every hour you spent reading split coverage, watching split explainer videos, or debating split implications on forums was an hour you didn’t spend reading NVIDIA’s supplemental disclosures.

The quarterly dividend increase from $0.04 to $0.10 per share was also embedded in that same May 22 announcement. As Hartford Funds has documented, splits rank among the most commonly misunderstood corporate actions precisely because they’re numerically simple — one number becomes ten, price adjusts, story writes itself. The $80 billion buyback requires you to model share count reduction across multiple quarters and translate that into EPS accretion. That’s harder, and harder doesn’t get clicked.

And because it’s harder, retail attention flows to the zero-value event while the high-value signal goes underanalyzed.

This asymmetry is precisely how institutional investors maintain their edge. Not through better data (the 10-Q is public), but through better allocation of analytical attention.

---

Stop doing this. NVIDIA has split its stock six times since 2000. Some splits preceded sustained runs. Some preceded drawdowns. The Nasdaq split calendar tracks these events across the market, and the data is noisy enough that no credible analyst I know builds a split into a forward model as a positive catalyst.

If you’re screening for companies that have recently split as a buy signal, you’re pattern-matching on a cosmetically altered number. You wouldn’t value a house differently because the owner subdivided the lot into ten parcels and kept all of them. Don’t do the equivalent with equity.

One boundary condition worth naming: this framework applies to forward splits of profitable, going-concern businesses. Reverse splits are a different instrument with different informational content — that’s a separate analysis for a separate issue.

I spent twenty-two years watching smart people misallocate their sharpest resource on the prettiest nothing. The split cost NVIDIA exactly $0 to execute. It cost you something far more expensive than money.

It cost you the hour you should have spent on the $80 billion.

That $80 billion is still working.

Your hour isn’t.

More where this came from