Solar TAM for AI-adjacent demand is driven by hyperscaler renewable energy procurement, US domestic manufacturing growth under IRA, and co-located solar deployments at AI campuses. The base case reflects continued hyperscaler PPA signing at current pace through 2028, First Solar expanding US capacity under IRA incentives, and co-located solar+BESS reaching commercial scale. Bear case assumes tariff policy reversal or grid interconnection backlogs delay solar deployments faster than new capacity comes online — reducing near-term installation. Bull case assumes nuclear SMR deployment is slower than expected, making solar+BESS the dominant AI data center power solution through 2030, and hyperscaler demand pulls forward GW-scale procurement ahead of current schedules.
“Hyperscalers are signing the largest solar power purchase agreements (PPAs) in history to power AI data centers. Microsoft, Google, Amazon, and Meta collectively contracted for 50+ GW of new renewable energy in 2025–2026, with solar representing ~60% of those commitments. AI data center power demand is the single largest driver of new US solar deployments.”
“First Solar — the only US-headquartered thin-film solar manufacturer at scale — is benefiting from Inflation Reduction Act domestic content bonuses and tariff protection on Chinese modules. First Solar guided record revenue and backlog for 2026, driven by hyperscaler and data center customers specifically requesting Made-in-America solar to qualify for IRA incentives.”
“AI data centers adopting co-located solar + battery storage (BESS) for 24/7 clean power commitments. Google DeepMind's AI-optimized grid management reducing solar curtailment. Microsoft piloting on-campus solar+storage at its AI campuses in Nevada and Texas. Co-location eliminates transmission losses and reduces grid interconnection queue risk.”
“Chinese solar panel manufacturers — LONGi, Jinko, Canadian Solar — face US tariffs of 25–50% on module imports and 3,000%+ on certain Chinese cells. This is creating a bifurcated market: domestic First Solar winning AI/hyperscaler orders; Chinese manufacturers competing in unsubsidized international and utility markets. Global module pricing near all-time lows on Chinese overcapacity.”
“Perovskite solar cells reaching commercial pilot stage with tandem efficiencies exceeding 33%. Oxford PV and First Solar partnering on next-generation high-efficiency panels targeting AI data center rooftop and parking deployments. Higher efficiency panels reduce land area required for on-campus solar — critical for data centers in land-constrained urban or suburban locations.”
Microsoft, Google, Amazon, and Meta have each committed to 100% clean energy for all data centers — including AI clusters. Solar is the fastest-deployable clean energy source, with 12–18 month development timelines vs. 5–10 years for nuclear or large hydro. Hyperscalers are collectively signing 50+ GW/year of renewable PPAs, with solar representing the majority. This is the largest corporate energy procurement wave in history, driven almost entirely by AI compute.
Hyperscaler new AI data center power demand by 2030: ~100 GW total
Solar share of that mix: ~50–60% = 50–60 GW new solar
Avg utility-scale solar cost: ~$0.8–1.2/W installed = $40–72B capex
Incremental above pre-AI baseline hyperscaler solar: ~$15–28B
→ First Solar, Enphase, inverter makers are primary beneficiaries
The Inflation Reduction Act provides a 10% domestic content bonus on top of the 30% investment tax credit for solar projects using US-made components. For AI data centers specifically, hyperscalers prefer domestic panels to satisfy ESG commitments and avoid tariff risk. First Solar — the only US thin-film manufacturer at scale — is sold out through 2028 entirely on domestic demand. IRA creates a structural, policy-backed demand floor that does not exist for Chinese manufacturers.
IRA domestic solar manufacturing incentives: $0.04/W production credit
US solar deployment 2026–2029 target: 80–100 GW
Domestic content premium captures: ~$3–4B in additional revenue vs. imports
First Solar capacity expansion: 14GW US capacity by 2026 (from 6GW in 2022)
14GW × $0.35/W ASP premium over Chinese modules = $4.9B additional revenue
→ Total IRA manufacturing tailwind: $8–14B cumulative 2026–2029
Hyperscalers are increasingly building solar arrays and battery energy storage directly on or adjacent to AI campuses — bypassing grid interconnection queues entirely. Amazon's Project Rainier in Washington, Microsoft's Nevada AI campus, and Google's Texas AI cluster all include on-campus solar+BESS. At scale, co-location reduces transmission losses, eliminates grid connection wait times (currently 5+ years), and provides energy cost certainty for 20-year campus lifecycles.
5 major AI campuses × 500 MW on-campus solar each = 2.5 GW
Co-located BESS: 2–4 hours storage × 500 MW = 1–2 GWh per campus
Solar: $0.9/W × 2.5 GW = $2.25B; BESS: $300/kWh × 10 GWh = $3B
Total across 5 campuses: ~$5–6B; scaling to 15 campuses by 2030
→ $5–10B net incremental solar+BESS at AI campuses above utility PPA baseline