Phase Space Research

NVIDIA

NVDA · Investment summary · as of 5 August 2026

Priced below what the business already delivers

Portfolio decision
Held in the growth book
Position size
5.1%
Price · 5 August 2026
$211.94
12-month target
$262.55 +24%
Expected return
Not determined
Next decision point
Date not announced

Business type: Cyclical · cyclical or commodity-sensitive

Size is the strategy's, set against its own position and exposure limits.

Investment view

At $211.94, NVDA requires a 19% five-year revenue growth rate to justify its enterprise value — less than the business already delivers, at 46%.

The disagreement with the market is about growth durability. The required terminal margin (43.26%) sits inside the demonstrated range on every post-2023 basis, while the required growth (40.30%) is the number doing the work.

The value rests on an exit multiple of 23.3x, a terminal operating margin of 54% and a 12.1% cost of capital. Move any one of them materially and the conclusion moves with it, which is why the required-versus-demonstrated test above carries more weight here than the point value.

The strongest argument against this view: One of the three direct customers making up 21%, 17% and 16% of Q1 FY2027 revenue re-phases or defers its FY2027-28 accelerator budget in favour of its own custom-silicon programme.

Underwriting bridge

QuestionEvidence-based conclusion
What drives the business?The disagreement with the market is about growth durability.
What do we forecast?Revenue growth of 46% demonstrated; a terminal operating margin of 54%; an exit multiple of 23.3x.
What does Street forecast?Not determined — no consensus estimates are joined to this record
Where do we differ?On revenue growth, the difference between what the price requires and what the business has demonstrated is +26.3 percentage points.
What is it worth?Twelve-month target $262.55, +24% from the struck price.
Why now?Date not announced — no dated event that would resolve the disagreement is on file

What must go right

  1. By Q4 FY2028 printFY2028 (year ending Jan-2028) revenue at or above $499bnWhere it stands: filed annual revenue
  2. By each 10-QAny one of the top three direct customers falls at or above 10% of revenue in two consecutive quartersWhere it stands: customer concentration disclosure
  3. By each 10-QCumulative inventory and purchase-obligation charges at or below $10bn across any four consecutive quarters, and inventory days at or below 180 at any quarter endWhere it stands: inventory charges; inventory days on TTM COGS

Catalysts and falsifiers

Date or windowEventThesis confirmed ifThesis weakened or refuted if
Q4 FY2028 printFiled annual revenueFY2028 (year ending Jan-2028) revenue at or above $499bnFY2028 (year ending Jan-2028) revenue below $499bn
each 10-QCustomer concentration disclosureAny one of the top three direct customers falls at or above 10% of revenue in two consecutive quartersAny one of the top three direct customers falls below 10% of revenue in two consecutive quarters
each 10-QInventory charges; inventory days on TTM COGSCumulative inventory and purchase-obligation charges at or below $10bn across any four consecutive quarters, and…Cumulative inventory and purchase-obligation charges above $10bn across any four consecutive quarters, OR inventory…
each 10-QGAAP operating marginGAAP operating margin at or above 43.3% for two consecutive quartersGAAP operating margin below 43.3% for two consecutive quarters

Risk and sell discipline

Impairment case

One of the three direct customers making up 21%, 17% and 16% of Q1 FY2027 revenue re-phases or defers its FY2027-28 accelerator budget in favour of its own custom-silicon programme. This is an impairment case, not a valuation bear case: NVIDIA is contractually obliged on ~$119bn of manufacturing/supply/capacity commitments against which NO customer is obliged to buy anything, and the H20 episode already demonstrated that this asymmetry converts a demand shock into a hard charge ($4.5bn of inventory and purchase-obligation charges in Q1 FY2026 on top of $4.0bn of FY2026 provisions). The exposure has grown: inventory days 93.2 -> 143.7 in twelve months, receivables +83.9%. And the revenue does not return - displaced accelerator revenue reappears on Broadcom's and Marvell's income statements, because a customer that has qualified its own ASIC does not requalify NVIDIA when the cycle turns.

Estimated probability 25%. No loss magnitude is on file, so the level at which the position would be resized cannot be struck and this case cannot be monitored against a loss budget.

Fundamental invalidation

Falsifiable and fundamental — not one of them is a price condition.

Price-based risk trigger

A daily close below $186.51 triggers an immediate review of the thesis and pauses additional buying. It is not an automatic sell unless a separate fundamental invalidation condition has been breached.

Upside sell discipline

No percentage return hurdle is used and none is introduced. On approach to the $262.55 target the case is reviewed rather than added to; a target reached is a reason to re-examine the position, not to hold it by default.

Investment criteria

CriteriaStatusInvestment meaning
QualityMetIs the business worth owning under its declared economic type?
ValuationMetIs the operating path required by today's price achievable?
LiquidityMetCan the intended position be built and exited in the right vehicle?
DownsideMetOne of the three direct customers making up 21%, 17% and 16% of Q1 FY2027 revenue re-phases or defers its FY2027-28 accelerator budget in favour of its own custom-silicon programme.
MomentumNot determinedDoes price action support or complicate entry timing? Not established on the evidence on file.
CatalystNot determinedIs there a dated event that resolves the disagreement? Not established on the evidence on file.
ConsensusNot determinedIs the house-versus-Street disagreement identified and quantified? Not established on the evidence on file.

Quality, valuation and liquidity can prevent a position on their own. The remaining four inform timing, sizing and monitoring, and never reject an investment by themselves.

Bottom line

The strongest case for mispricing is that the business already delivers +26.3 percentage points more growth than the price requires. The most important unresolved uncertainty is the permanent-loss mechanism: one of the three direct customers making up 21%, 17% and 16% of Q1 FY2027 revenue re-phases or defers its FY2027-28 accelerator budget in favour of its own custom-silicon programme. The next evidence that should change the portfolio decision is the next scheduled results, or a daily close below $186.51, which forces an immediate review.