2026年一季度AI公共报表和估值指南（英）2026

EMERGING TECH RESEARCH AI Public CompSheet andValuation Guide Key takeaways Institutional Research Group Dimitri ZabelinSenior Research Analyst,AI and Cybersecuritydimitri.zabelin@pitchbook.com •AI semiconductors fell by 2% on a weighted basis in Q1 2026—the only cohort to hold near flat and outperform the S&P 500 (-4.6%).Every other AI cohort underperformed by a wide margin: AI vertical conglomerates fell by a mean of 37%, AI core pure plays declined31.8%, and AI core conglomerates fell 14.9%. Oscar AllawaySenior Data Analystpbinstitutionalresearch@pitchbook.comPublished on April 22, 2026 •The uniform thematic premium that supported elevated multiples from 2023 to 2024 has reversed. Capital is concentrating incompanies where AI spending is directly reflected in revenue growth and margin expansion, and it is moving away from companieswhere AI remains an embedded feature without a defined monetization pathway. Contents Key takeaways2Stock returns7Revenue8EBITDA10 •AI vertical conglomerates delivered 10% to 21% revenue growth in fiscal year 2025 yet saw enterprise value (EV)/trailing 12-month(TTM) revenue compress to approximately 4.5x on a weighted basis—the lowest level in this dataset. The market has stoppedrewarding AI feature integration that does not produce measurable net-new monetization. •Companies with expanding EBITDA margins—Palantir at 51%, AppLovin at 82.3%, and NVIDIA at 67.5%—held up relative to peers.Names where margin lagged revenue growth, including Snowflake (6% EBITDA margin), MongoDB (14.9% EBITDA margin), and Upstart(no EBITDA), declined more than 40% YTD. •Chinese AI-linked names underperformed Western peers independent of operating results. Alibaba declined 17.2%, Baidu 20.1%, andTencent 19.8%, while Cambricon declined 26.6% despite 452.1% revenue growth in fiscal-year 2025, reflecting sustained geopoliticaland liquidity discounts. PitchBook clients can access thefull Excel data pack for this reportvia theResearch Centeron thePitchBook Platform. •The first is the pace at which software companies convert AI investment into measurable revenue in the Q1 2026 to Q2 2026 earningscycles. The second variable is the durability of semiconductor demand: NVIDIA guided for sequential revenue growth throughoutcalendar 2026 with shipments extending into 2027, supported by hyperscaler capital expenditure commitments estimated atapproximately $527 billion for 2026. Stock returns AI core conglomerates declined by a mean of 14.9% and a weighted average of 14.7% in Q1 2026, underperforming the S&P 500 byapproximately 10 percentage points. CoreWeave was the sole positive performer at 8.2%, reflecting concentrated infrastructure exposure.Oracle led declines at 24.5%, followed by Microsoft at 23.5% and Baidu at 20.1%. Notably, strong fiscal-year 2025 results—Microsoft with18% and Meta with 21.2% EBITDA growth—provided no insulation, indicating a structural discount applied to diversified AI exposure. On aone-year basis, CoreWeave rose 108.9% versus a cohort median of 5.2%. AI core pure plays declined by a mean of 31.8%, with AppLovin, MongoDB, and Upstart each down more than 40%. Palantir was relatively resilientat a 17.7% decline, supported by reported profitability. The selloff reflects a partial reversion from elevated 2025 entry points—Palantir had risen73.3% and AppLovin 50.2% on a one-year basis—compounded by a market now applying a higher evidentiary standard to margin validation. AI vertical conglomerates was the worst-performing cohort, declining by a mean of 37%. Atlassian fell 57.9%, The Trade Desk 40.2%,Workday 39.5%, and HubSpot 39.2%. EV/revenue compressed to a weighted average of approximately 4.5x despite steady top-line growth.Until AI demonstrably drives higher net revenue retention rather than primarily reducing churn, investors are unwilling to assign a premiumabove conventional software-as-a-service multiples. AI semiconductors fell 2% on a weighted basis. Standouts included Arm Holdings (up 38.4%), ASML (up 23.5%), Micron (up 18.4%),SK Hynix (up 17.4%), and Marvell (up 16.6%). NVIDIA declined 6.5%, and Astera Labs fell 34.1% due to a CFO departure and valuationcompression after a steep run-up. Despite 115.1% fiscal-year 2025 revenue growth, the decline illustrates that even infrastructure namesface valuation risk when expectations are set at a premium. AI autonomous machines declined by a mean of 12.6%. Aurora Innovation was the only positive performer at 7.3%, following itscommercial autonomous trucking launch in April 2025. UiPath declined 32.3% as agentic AI frameworks from foundation model providersincreasingly overlapped with its core robotic process automation use cases. Mobileye fell 34.2% on automotive original equipmentmanufacturers softness and inventory correction rather than technology deterioration. Revenue Fiscal-year 2025 results reinforce the structural layering of the AI stack, with growth rates declining progressively from semiconductors todiver