The Best AI Stocks To Buy In 2026

Finance-focused Summary (AI “Supercycle” Stock Watchlist + Framework)

Macro / Thesis

• The speaker argues the AI cycle is not finished and is evolving into a broader agentic AI supercycle.
• Claim: Smart investors will benefit while most investors will underperform.
• Core argument: agentic AI needs massively more infrastructure, including:
  • compute
  • memory
  • networking
  • power
  • cybersecurity
  • orchestration
• Demand acceleration mechanism:
  • As costs drop, the speaker expects faster adoption via a Jevons paradox / flywheel effect.
• The speaker references Jensen Huang, claiming the agentic AI inflection point is already here (as referenced in Nvidia’s Q4 earnings, per speaker).

Methodology / Investing Framework

Two-part requirement before buying

1. Invest in the story/potential.
2. Only buy if fundamentals already exist (i.e., “proof of concept”).

AI stack layers to focus on (infrastructure > applications)

• Layer 1: Compute
  • chips/engines, memory
  • EDA
  • semiconductor equipment
• Layer 2: Cloud & Power
  • cloud providers
  • cooling
  • grid upgrades
  • onsite power
  • optical networking
  • data-center real estate
• Layer 3: Orchestration
  • data
  • governance/monitoring
  • cybersecurity
• Layer 4: Embodied AI / Robotics
  • physical AI / robotics

Position sizing system: DCA + trimming (BCAX2)

• Deploy 50% / 50% initially, then wait.
• If the stock comes down, deploy double.
  • Next tranche totals 150% deployed.
• Adjust when price is above/below thresholds.
• Trim rules (then reinvest proceeds into continued DCA):
  • trim when stock is up about: +50%, +100%, +150%, +200%

Risk management rules

• Emergency fund: keep 6 months of expenses before investing.
• “Never sell” based purely on:
  • the stock going up
  • the stock going down
  • macro changes
• Use trim + redeploy via the system rather than reactive selling.

Diversification / disciplines cited

• S&P exposure is “critical”:
  • speaker claims at least half of the portfolio in the S&P.
• Bonds and DCA emphasized.

Disclaimers / cautioning language

• Noted: promotional tone (e.g., “everything free/no paywall”) and no explicit “not financial advice” disclaimer in the subtitles.

---

Key Numbers and Fundamental Claims Cited (by Company)

Layer 1: Compute (Semiconductors, Memory, Data-center Networking, EDA, Tools)

Nvidia (NVDA)

• Revenue: around ~$220B, growing ~65%/yr
• Gross margin ~75%
• Operating income: growing ~60%/yr
• Valuation vs growth (example claims):
  • earnings “blew past” prior multiple
  • revenue progression referenced: 60 → 220 (speaker also cited intermediate points “61, 131, 216”)
  • stock performance: “5x” from late 2023 to “now” (as described)
• Strategic moat:
  • “ecosystem lock-in”
  • multiple capex cycles ahead (5–6 cycles claimed)

Broadcom (AVGO)

• Described as custom AI chips + networking
• AI semiconductor YoY metrics:
  • Revenue +143%
  • Operating income +89%
  • Net income +292%

AMD (AMD)

• “Second sourcing” of GPUs + major CPU exposure (framed as relevant as agentic AI shifts toward CPU/memory/context)
• YoY metrics:
  • Net income +164%
  • Operating income +94%
  • Free cash flow +180%

ARM (ARM)

• “Blueprint” + royalties “tollbooth model”
• No specific numeric results provided beyond qualitative linkage to royalty growth.

ASML (ASML)

• Lithography equipment near-monopoly (speaker: “only one company”)
• Claims:
  • booked for the next 5 years at full capacity (capacity constraint narrative)

Micron (MU)

• Shift from commodity to moat via scale + high-bandwidth memory
• Claims:
  • trading around 5x forward earnings (despite price spike)
  • Operating income +650% (past year)
  • Free cash flow +1,300% (past year)
  • “Beginning” claim for continued upside

Arista Networks (ANET)

• Data center switching/networking to interconnect GPUs
• Claims:
  • Revenue growing ~30%
  • Income growing ~23%
  • Earlier: up ~340% since being added in Oct 2020 (per speaker)

Cadence Design Systems (CDNS)

• EDA software to simulate chip design/production flow and prevent costly errors
• Claims:
  • Free cash flow growth ~42%
  • Operating income growth ~22%
  • Operating margin ~31%

Lam Research (LRCX)

• Chip etching equipment; described as ~50% market share
• Claims:
  • Net income +40% YoY
  • Revenue +24%
  • Operating income +40% YoY

TSMC (TSM)

• Manufacturing near-monopoly narrative
• Geopolitical risk noted:
  • “Taiwan…disputed/potentially geopolitically problematic”
• Claims:
  • Operating income growth ~48%
  • trading below 20 forward P/E (valuation point)
  • “Sold out into infinity” (capacity narrative)

---

Layer 2: Cloud & Power (AWS/Azure/Google; Cooling; Grid; Onsite Power; DC Infrastructure)

Amazon (AWS) (AMZN)

• Described as a “toll booth” AI cloud
• Claims:
  • Net income growth 31% (past year)
  • Valuation: 3.6 price-to-sales
  • Valuation: 25 forward P/E

Microsoft (MSFT)

• Cloud + enterprise software/tools for running agents
• Claims:
  • ~20x earnings
  • ~47% operating margin
  • Valuation: “buy at 9x sales”
• Growth:
  • Cloud growth ~63% annual growth (speaker: “fastest cloud growing”)
  • Net income growth ~32% YoY
• Valuation point:
  • ~23x forward earnings (speaker: “good price”)

Google (GOOGL/GOOG)

• TPUs described as improving margins vs reliance on pure GPU
• Claims:
  • Cloud growth ~63% (speaker links to “fastest cloud growing,” in Google comparison context)
• No explicit TPU unit economics numbers provided beyond the margins narrative.

Vertiv (VRT)

• Cooling/thermal management
• Claims:
  • Net income +170%
  • Free cash flow +67%
  • Operating income +38% (past year)

GE Vernova (GEV)

• Grid hardware/turbines required for data center expansion
• Claims (past year):
  • Operating income +76%
  • Net income +214%
  • Operating profit margin growth +62%

Ciena (CIEN)

• Optical networking connecting data centers
• Claims:
  • Income growth 47%
  • Operating income 63%
  • Free cash flow 76% (past year)

Bloom Energy (BE)

• Onsite power (mitigating grid strain)
• Mentions skepticism (“might be just lying” caveat), then provides bullish figures
• Claims:
  • Revenue ~+40% YoY
  • Operating income +217% YoY
  • Free cash flow +72% YoY

Constellation Energy (CEG)

• Nuclear for 24/7 reliable power
• Claims:
  • Trading at ~3x sales
  • Free cash flow growth +125% YoY
  • Emphasis: reliability + non-speculative framing

Digital Realty (DLR)

• Data-center real estate “landlord”
• Claims:
  • Net income +117%
  • Free cash flow +520%
  • EBITDA margin ~61%

---

Layer 3: Orchestration (Governance, Data, Security)

Palantir (PLTR)

• Speaker: “top conviction”
• Claims:
  • Revenue growth 85%
  • “Rule of 40” = 145 (framed as “four times highest than [the] daily number”)
  • Operating income growth +360% YoY
• Valuation contrast (speaker narrative):
  • “17x sales in 2023,” revenues doubled, price went to “20x”
  • later claim: ~58x sales at that point (used to argue fundamentals caught up)

MongoDB (MDB)

• Database backbone for model output/context
• Claims:
  • Free cash flow growth +314% (last year)
  • Revenue growth +23%
• Qualitative: “context bottleneck”

CrowdStrike (CRWD)

• Cybersecurity enabling independent agent access to sensitive data/code
• Claims:
  • Net dollar retention 120%
  • Revenue +22%
  • Free cash flow +23%

“AI net”

• Mentioned as a company in layer one and layer three.
• No specific ticker provided in the subtitles.

---

Layer 4: Embodied AI / Robotics (Robotics decade thesis; Tesla lead)

Tesla (TSLA)

• Framed as poised to dominate robotics by 2035
• References:
  • Optimus + FSD as robotics products
  • “FSD is a robot…wheels instead of arms/legs”
• Claims (financial/context):
  • Free cash flow growth +74% (past year)
  • Cash: $44B
  • Annual revenue: ~$95B from selling cars
• Strategic argument:
  • Tesla’s sensors/vision + motors/batteries + supply chain + manufacturing scale
  • “10 million real miles” data collection narrative
• Skeptic framing:
  • lidar vs cameras (Elon Musk cited: cameras cheaper and integrated with existing robotics efforts)

---

Presenters / Sources (As Stated)

• Presenter: Don Nash (referred to as “Tom” in one moment; later mentions “Patreon…Roth Academy…don-nash” and directs viewers to “pEkingnology.com/don-nash”).
• Named source cited: Jensen Huang
  • referenced as stating the agentic AI inflection point is already here (per speaker, tied to Nvidia’s Q4 earnings).