Which AI stocks benefited most since ChatGPT launched in November 2022?

As of June 2026, the largest approximate total returns since ChatGPT's launch (Nov 30, 2022) include Micron (~+1,761%), NVIDIA (~+1,136%), Broadcom (~+620%), Vistra (~+500%), Meta (~+397%), and Constellation Energy (~+317%). The S&P 500 returned ~+65% over the same period — roughly half of which was driven by the top 5 mega-cap AI names.

How much AI exposure is already in the S&P 500?

As of mid-2026, the top 5–6 mega-cap technology companies (NVIDIA, Microsoft, Apple, Alphabet, Amazon, Meta) represent approximately 28–32% of the S&P 500 by market cap weight. A standard S&P 500 index fund is therefore meaningfully concentrated in AI-adjacent names, even without any active investment decision.

Insights — Q2 2026

AI Stocks: Who Actually Won —
And What Comes Next

Q: Are AI stocks overvalued in 2026?

It depends on the wave. Wave 1 (compute/GPU) names like NVIDIA are priced for near-perfection and have limited re-rating runway. Wave 2 (HBM memory) and Wave 3 (power/energy infrastructure) still have measurable demand tailwinds — HBM demand is projected to grow 2–3× by 2027 vs. 2024, and data center power demand is expected to double by 2030. Wave 4 (agentic applications) is nascent and high-variance. 'AI stocks' is not a monolithic category.

Q: Should RSU holders at tech companies diversify out of AI stocks?

For most high earners at tech companies, the combination of unvested RSUs, an S&P 500 index fund in their 401(k), and any individual AI stock holdings creates a 'triple overlap' — the same AI names appear three times in their net worth. When your career income and your portfolio are correlated to the same sector, diversification isn't conservative — it's rational hedging. The main friction is tax efficiency: unwinding concentrated positions with low cost basis triggers capital gains, so strategies like exchange funds, charitable vehicles (DAFs), or systematic sales over multiple tax years are worth modeling.

Q: What is the next AI investment wave after compute and GPUs?

The investment community has broadly identified three waves following the initial GPU/compute boom: (1) Memory infrastructure — specifically High Bandwidth Memory (HBM), where SK Hynix and Micron are the primary beneficiaries of a 2–3× demand increase projected through 2027. (2) Power and energy infrastructure — data center electricity demand is expected to double by 2030, making nuclear power operators and grid infrastructure companies like Vistra and Constellation structurally attractive. (3) Agentic application layer — vertical SaaS companies that replace entire workflows (rather than augmenting them) and can charge on outcomes rather than seats.

The AI rally since November 2022 made a handful of stocks legendary. But for high earners, the real question isn't who won — it's whether you already own too much of the same names through your 401(k), index funds, and RSUs. A breakdown of the four investment waves, a sector performance heatmap, and a free quiz to identify your investor archetype.

Last updated June 9, 2026 · ~10 min read · Not financial advice. See disclaimer.

+1,136%

NVDA gain since ChatGPT launch (Nov '22 to Jun '26)

StatMuse / public market data (split-adjusted)

~30%

S&P 500 weight in just 5 mega-cap AI names (Jun 2026)

S&P Dow Jones Indices

~$700B

Combined AI capex committed by 4 hyperscalers in 2026 (Amazon, Google, Microsoft, Meta)

Q1 2026 earnings calls; Financial Times compilation

2–3×

HBM memory demand growth projected by 2027 vs. 2024 baseline

IDC / TrendForce estimates, 2026

Nov 2022 → Jun 2026

The "ChatGPT Rally" — Who Actually Won

ChatGPT launched November 30, 2022. In the 43 months since, the gap between AI winners and the broader market has been historic — but unevenly distributed. Here's the approximate total return by name, sorted by magnitude.¹

Approximate Total Return — AI-Adjacent Equities vs. Benchmarks

Nov 30, 2022 → Jun 2026. All figures approximate; actual returns vary based on exact purchase/sale dates and dividends. Not investment advice.

+500% or more

+100–500%

+50–100%

Benchmark / Laggard

Sector × Performance

AI Stock Heatmap by Wave & Sector

Not every "AI stock" is the same bet. The rally has moved in distinct waves tied to the physical and software infrastructure of AI. Each wave has its own risk profile and forward setup.

Company	Wave	Est. Return Nov'22–Jun'26	Current Momentum	Key Thesis
NVIDIA (NVDA) GPU / Compute	⚡ Compute	~+1,136%	→ Digesting	Dominant GPU platform; Blackwell cycle in full swing; valuation pricing in perfection
Meta (META) AI-driven Ads / LLaMA	🤖 Apps	~+397%	↑ Strong	AI ad-targeting monetization + open-source model strategy; capex cycle ramping
Broadcom (AVGO) Custom AI ASICs / Networking	⚡ Compute	~+620%	↑ Strong	Custom ASIC (XPU) demand from hyperscalers; networking silicon beneficiary
Microsoft (MSFT) Cloud AI / Copilot	☁️ Software	~+66%	→ Market	Azure AI growth; Copilot monetization ramping; OpenAI partnership structural
SK Hynix (000660.KS) HBM Memory	🧠 Memory	>+2,000%	↑ Strong	Primary HBM3E supplier for NVDA H200/B100; return in KRW local currency terms. Note: Micron (MU) has outpaced SK Hynix in USD total return (~+1,761%) as its HBM ramp accelerated and it received NVIDIA HBM4 certification.
Micron (MU) HBM / DRAM	🧠 Memory	~+1,761%	↑ Strong	HBM ramp now matching and challenging SK Hynix; DRAM cycle + HBM3E certification by NVIDIA has driven ~+1,761% from Nov 2022. Massive re-rating as AI memory demand became clear
TSMC (TSM) Foundry / N3/N2	⚡ Compute	~+157%	↑ Strong	Every AI chip is fabbed here; N2 ramp in 2026; geopolitical risk priced in partially
Vistra (VST) Power / Nuclear	⚡ Energy	~+500%	↑ Strong	Data center power demand driving nuclear repricing; 24/7 clean power contracts. Note: Constellation (CEG) returned ~+317% over same period.
Alphabet (GOOGL) Search AI / TPUs / Cloud	☁️ Software	~+289%	↑ Strong	Gemini/search integration; GCP AI growth; margin expansion offsetting disruption concerns
AMD GPU / AI Accelerators	⚡ Compute	~+80%	↓ Lagging	MI300X momentum; still a distant #2 to NVDA in AI GPU share; software moat challenge
S&P 500 (SPY) Benchmark	—	~+65%	→ Market	Note: ~30% of this return is itself driven by the 5 mega-cap AI names

✦ The Benchmark Trap

The S&P 500's ~65% gain since Nov 2022 looks healthy — but roughly 30% of the index weight itself sits in NVDA, META, MSFT, GOOGL, and AMZN, per S&P Dow Jones Indices, meaning these five names drove a disproportionate share of that return.⁶ If you hold a standard index fund and work at one of these companies, you are not diversified. You are concentrated.

Investment Thesis Framework

The Four Waves: What Fueled Each Move

Every technology platform build-out follows a similar infrastructure-to-application arc. AI is no different — but the waves have moved faster than any prior cycle. Understanding where each wave stands tells you where capital is likely to rotate next.

Wave One · Largely Played

Compute — "Picks and Shovels for the Gold Rush"

Nov 2022 → Mid 2024

The first and most explosive wave was straightforward: every AI model needs GPUs to train. NVIDIA was the only vendor with the full hardware-software stack (CUDA ecosystem) that research teams had spent a decade building on. Hyperscalers and startups raced to secure H100 allocation, creating a supply/demand shock that drove NVDA's data center revenue up more than 3× in a single fiscal year (FY2023→FY2024).²

NVDA ~+890%

AVGO ~+450%

TSMC ~+180%

ARM ~+120%

✓ Largely priced in at current valuations

Wave Two · Active Now

Memory — The Current Physical Bottleneck

Mid 2024 → 2027 (estimated)

AI models don't just need compute — they need to move data to and from the GPU faster than traditional DRAM allows. High Bandwidth Memory (HBM) stacked directly on the GPU die is the only solution. SK Hynix is currently the primary supplier of HBM3E used in NVIDIA's H200 and B100 series. Micron has qualified as a second source on select SKUs; Samsung is ramping but lags both. IDC projects HBM demand growing 2–3× by 2027 vs. the 2024 baseline, against constrained supply expansion due to complex packaging yields.³

SK Hynix Active

MU Leading

Samsung Lagging

▶ Actively playing out — forward runway remains

Wave Three · Early Innings

Power & Energy Infra — The Grid is the New Bottleneck

2025 → 2028+

A single H100 GPU cluster draws as much electricity as a small town. The four major hyperscalers (Amazon, Google, Microsoft, Meta) have collectively committed ~$700B in capex for 2026 — up 77% from 2025's already-record $410B — and data center power demand is projected to double by 2030.⁴ The constraint is no longer chips — it's getting reliable, 24/7 clean electricity to new data centers. Nuclear power plants (particularly existing ones) became the most sought-after real estate in energy. Vistra and Constellation re-rated dramatically when Microsoft and Google signed long-term nuclear power purchase agreements.

VST ~+500%

CEG ~+317%

GEV Active

EATON Active

◎ Early innings — power demand secular, not cyclical

Wave Four · Nascent

Agentic Application Layer — Where Compounding Happens

2026 → 2030+

Waves 1–3 are infrastructure. Wave 4 is where the infrastructure starts earning. AI agents — systems that can take multi-step actions autonomously — are beginning to replace workflows rather than just accelerate them. The early winners are vertical SaaS companies that have embedded AI deeply enough to charge on outcomes rather than seats. The transition from "training" capex to "inference" at the edge will shift revenue from chip companies toward platform companies. This is the wave that's hardest to call but has the longest compounding runway — and where most of the identifiable public companies are still early in their S-curves.⁵

PLTR Early

ServiceNow Early

Salesforce Early

Vertical SaaS Nascent

◈ Nascent — hardest to time, longest duration

✦ The Rotation Question

Capital doesn't disappear from tech — it rotates. As NVDA multiple expansion slows, the marginal dollar has moved toward memory, power, and agentic applications. Sophisticated investors aren't asking "is the AI trade over?" — they're asking "which wave am I in?"

Concentration Risk

Portfolio Checkup: Your Hidden AI Exposure

For high earners, the AI rally may not be an opportunity — it may already be a risk you're not accounting for. Check how many of these apply to you.

Interactive Checkup

How Concentrated Are You, Really?

Select every statement that applies to your situation. Your concentration profile will update in real time.

I work at a Big Tech or AI company and hold unvested RSUs

Your human capital and equity are correlated with the same AI narrative. A single sector downturn hits income and net worth simultaneously.

My 401(k) is primarily in a target-date fund or S&P 500 index

As of June 2026, NVDA + MSFT + META + GOOGL + AMZN represent roughly 28–32% of the S&P 500 by weight. Your "diversified" fund is heavily concentrated in the same AI names.

I hold individual AI stocks (NVDA, META, MSFT, GOOGL, etc.) outside of index funds

Intentional concentration can be appropriate — but compounded with index exposure and RSUs, the same names appear three times in your net worth.

More than 25% of my liquid net worth is in a single equity position

A commonly cited threshold for concentration risk. At 25%+, the position's volatility meaningfully drives your total net worth outcomes regardless of other holdings.

I haven't reviewed my actual sector weights across all accounts in the past 12 months

With the market cap shift in AI names, a "balanced" portfolio from 2022 may now be significantly overweight tech without any active decision being made.

The "Double Dip" Problem

You may own NVDA three times without knowing it

RSUs at your employer → correlated to NVDA's supply chain. Your S&P 500 index fund → ~7% NVDA weight. Your sector ETF or individual buys → direct exposure. Stress-testing your net worth against a 40% AI sector drawdown is not paranoia — it's portfolio hygiene.

The Hedge Framing

Career risk and portfolio risk should move in opposite directions

If an AI downturn would both threaten your job and crater your portfolio, you're not hedged — you're leveraged. The sophisticate's move is intentionally diversifying into assets that do well when tech does poorly: value, international, commodities, real assets.

The Tax Angle

Concentrated positions have a tax cost to unwind — which is why most people don't

If you have a large NVDA or MSFT position with a low cost basis, selling to diversify triggers capital gains. The rational framing: compare the expected after-tax drag of diversifying now versus the expected downside of staying concentrated. Tools like exchange funds, charitable giving (DAF), or installment sales can help manage the transition. This is a calculation worth running with an advisor — not a reason to avoid it indefinitely.

Free · 5 Questions · 2 Minutes

Which AI Investor Type Are You?

Your career context, existing exposure, and risk tolerance define which part of the AI investment story is actually relevant to you. Take the quiz to identify your archetype and get a tailored read on the next wave.

AI Investor Type Quiz

Find Your Investor Archetype

Five questions. Honest answers. Your result maps to one of three investor archetypes — each with a different forward thesis for the AI trade.

0 / 5

1 of 5 — What's your primary source of equity exposure to AI right now?

Company RSUs at a tech or AI firm — this is my largest equity position

Passive index funds (S&P 500, total market) — AI is baked in but not intentional

Individual AI stocks I actively chose to buy

Minimal — mostly cash, bonds, real estate, or non-tech equities

2 of 5 — How do you think about the AI trade at current valuations?

It's overextended — I'm more concerned about downside than upside from here

Selectively interesting — some waves still have runway, others are priced for perfection

Secular, multi-decade shift — short-term volatility is noise, I'm adding on dips

Agnostic — I don't try to time sectors, I let index funds do their thing

3 of 5 — What does your career trajectory look like relative to AI?

Deep insider — I work in AI/ML, my career is directly tied to this industry

Adjacent — I work in tech or finance, AI is reshaping my work but I'm not building it

Power user — AI tools meaningfully change my productivity but I'm outside tech

Unrelated — my career isn't materially affected by AI either way yet

4 of 5 — What's your biggest concern with your current AI-adjacent portfolio?

Too concentrated — RSUs + index + individual stocks all point the same direction

Underexposed — I missed the first wave and I'm figuring out how to participate now

Timing — I believe in the thesis but I'm worried about buying near the peak

Tax drag — I have gains I don't want to trigger but the concentration bothers me

5 of 5 — Which "next wave" thesis resonates most with you?

Memory & hardware — HBM is a physical bottleneck, SK Hynix and Micron have runway

Power & energy infra — the grid is the new GPU, nuclear and grid stocks are mispriced

Agentic apps — vertical SaaS companies that replace workflows, not just assist them

Diversify away — my job already gives me AI upside, my portfolio should hedge it

Your Archetype

–

Frequently Asked Questions

The questions most high earners ask when they start thinking seriously about AI exposure in their portfolio.

Are AI stocks overvalued in 2026?

It depends on the wave. Wave 1 compute names like NVIDIA are priced for near-perfection with limited re-rating runway. Wave 2 (HBM memory) and Wave 3 (power infrastructure) still have measurable demand tailwinds — HBM demand projected 2–3× by 2027 vs. 2024, and data center electricity demand expected to double by 2030. Wave 4 (agentic applications) is nascent and high-variance. "AI stocks" is not a monolithic category, and the answer changes significantly depending on which wave you're asking about.

Which AI stocks benefited most since ChatGPT launched?

As of June 2026, the largest approximate total returns since ChatGPT's launch (Nov 30, 2022) include Micron (~+1,761%), NVIDIA (~+1,136%), Broadcom (~+620%), Vistra (~+500%), Meta (~+397%), and Alphabet (~+289%). The S&P 500 returned ~+65% over the same period — roughly half of which was itself driven by the top 5 mega-cap AI names. All figures are approximate and vary based on exact dates used.

How much AI exposure is already inside the S&P 500?

As of mid-2026, the top 5–6 mega-cap technology companies (NVIDIA, Microsoft, Apple, Alphabet, Amazon, Meta) represent approximately 28–32% of the S&P 500 by market cap weight, per S&P Dow Jones Indices. A standard index fund is therefore meaningfully concentrated in AI-adjacent names without any active decision being made. This weight has roughly doubled from where it sat in 2019.

Should tech RSU holders diversify out of AI stocks?

For most high earners at tech companies, unvested RSUs plus an S&P 500 index fund plus any individual AI holdings creates a "triple overlap" — the same names appear three times in their net worth. When your career income and your portfolio are correlated to the same sector, diversification isn't conservative — it's rational hedging. The main friction is tax efficiency: unwinding concentrated positions with low cost basis triggers capital gains. Strategies like exchange funds, donor-advised funds, or systematic annual sales are worth modeling with a financial advisor.

What is the next AI investment wave after GPUs?

Three waves follow the initial compute/GPU boom. Wave 2 is memory infrastructure — specifically High Bandwidth Memory (HBM), where SK Hynix and Micron are primary beneficiaries of a projected 2–3× demand increase by 2027 vs. 2024. Wave 3 is power and energy infrastructure — data center electricity demand is expected to double by 2030, making nuclear operators and grid companies like Vistra and Constellation structurally interesting. Wave 4 is the agentic application layer — vertical SaaS businesses that replace workflows (not just assist them) and can charge on outcomes rather than seats. This is the highest-duration, hardest-to-time wave.

References

Sources & Methodology

1Bloomberg Terminal / Yahoo Finance / StatMuse public price data. Return figures are approximate total returns, Nov 30, 2022 to Jun 9, 2026. Not investment advice.
2NVIDIA Corporation, Quarterly Earnings Reports Q4 FY2023 through Q4 FY2025. investor.nvidia.com. Revenue figures and YoY growth rates.
3IDC Worldwide Semiconductor Tracker, H2 2025 and H1 2026. TrendForce HBM supply/demand analysis, Q1 2026. HBM demand growth projections.
4Goldman Sachs Global Investment Research, "AI Data Centers and the Coming US Power Demand Surge," May 2024. Q1 2026 earnings calls: Amazon ($200B), Alphabet ($175–185B), Microsoft ($190B), Meta ($125–145B) — combined ~$700–725B, per Financial Times compilation, April 2026. IEA data center power demand projections, 2026.
5a16z "The AI Agentic Era," March 2026. Gartner Hype Cycle for AI, 2025. Sequoia Capital AI market analysis, Q4 2025.
6S&P Dow Jones Indices, index constituent weights as of May 2026. MSCI sector weighting methodology.

Written by

HenryPulse Research & Editorial Team

HenryPulse produces data-driven financial content for high-income professionals in tech, finance, and strategy. Our research combines public market data, earnings disclosures, and third-party industry reports.

Not financial advice. This article is for informational and educational purposes only. It does not constitute financial, tax, investment, or legal advice. Always consult a qualified professional before making financial decisions. Full disclaimer →

AI Stocks: Who Actually Won —And What Comes Next