| AI category | 2026 market size | YoY growth | Key players |
|---|
| Foundation model API | $30B | +200% | OpenAI, Anthropic, Google, Meta |
| AI infrastructure (GPUs, data centers) | $400B | +80% | NVIDIA, AMD, Broadcom, hyperscalers |
| AI applications (vertical SaaS) | $80B | +150% | Salesforce, ServiceNow, startups |
| AI services (consulting, integration) | $50B | +100% | Accenture, McKinsey, Big 4 |
| AI chips (custom silicon) | $80B | +120% | Google TPU, AWS Trainium, Microsoft Maia, Meta MTIA |
The $640 billion total is the operative macro signal. The $400 billion AI infrastructure line is NVIDIA-led, and NVIDIA's 56 percent data center growth confirms the line is real. The $30 billion foundation model API line at 200 percent growth is the new frontier; OpenAI, Anthropic, Google, Meta are the leaders, and the line is the highest-growth segment of the AI market. The $80 billion AI applications line at 150 percent growth is the SaaS layer that will absorb AI capabilities over 2027-2028.
Risks to the Mag 7 rally
Three risks concentrate in the next 12 to 18 months per the 2026 Stanford AI Index regulatory analysis. First, AI bubble concerns: AI capex commitments have grown from $200 billion in 2025 to $300 billion in 2026, and any sign of demand-side weakness (capex digestion, AI rollouts slow) would compress the rally. Second, hyperscaler capex sustainability: the four hyperscalers cannot sustain 30 percent capex growth indefinitely; the 2027 capex growth is the operative signal for the Mag 7 rally's durability. Third, regulatory risk: the EU AI Act (effective August 2026), the US Executive Order 14110 follow-on legislation, and any state-level AI law could impose compliance costs that affect Mag 7 economics.
How to invest in the Mag 7
The 2026 Mag 7 investment paths run through four vehicles per major broker platform research. Direct stocks give the most control and the most concentration risk. The Roundhill MAGS ETF gives equal-weight exposure across the basket with lower concentration risk. Sector ETFs (XLK tech, SMH semiconductors) give broader tech exposure without Mag 7 concentration. Thematic ETFs (AIQ, ROBO, BOTZ) give AI-specific exposure with basket diversification.
The 60-30-10 allocation framework is the operative allocation: 60 percent core diversified exposure (MAGS or XLK), 30 percent AI-specific exposure (AIQ or ROBO), 10 percent NVIDIA-specific exposure (NVDA direct) for the highest-conviction AI thesis. The 6 to 12 month time horizon is the right investment window; the AI trade is durable but rate-cycle-sensitive, and the 2027 capex reports will determine the next leg. Tax-efficient wrappers (IRA, Roth) are the right vehicle for retail investors with the long time horizon.
What to watch next
Four signals settle the rest of 2026 and 2027. NVIDIA Q3 FY2026 earnings on November 19 2026 is the first major catalyst; the data center revenue growth rate is the operative number. Hugging Face integration milestones (NVIDIA-HF product launch expected Q1 2027) will tell whether the $12.93 billion acquisition was accretive. The Mag 7 Q3 earnings calendar (October through late November 2026) will tell which companies can sustain the AI capex commitments. And the Fed rate cycle: a December 2026 rate cut would lift the growth-stock basket; a hold or hike would compress it.
Resources and next steps
- Read the Roundhill MAGS ETF page on roundhillinvestments.com for the daily NAV and equal-weight basket composition.
- Pull the NVIDIA Q2 FY2026 earnings release from investor.nvidia.com for the canonical data center revenue, gross margin, and FY2028 guidance.
- For US investors evaluating the 2026 Mag 7 AI trade: the Sep 3 MAGS ETF level is the consolidated read, NVIDIA's +45 percent YTD is the structural engine, the AI capex commitments are the supply-side commitment, and the enterprise AI adoption at 65 percent is the demand-side pull.
- For the silicon-side and the capex-side context, see our guides on the AI chip war landscape, the Amazon $17.5 billion credit line and $364 billion AWS backlog, the OpenAI NVIDIA Ohio data center lease, and the NVIDIA $4.5 trillion market-cap analysis.