2026年上半年市场状况

February 2026 3Letter From the Authors5Perspectives on the Innovation Economy6Macro11Fundraising and Investment20Regional Tech Hubs24AI Adoption and Investment27VC-Backed Company Benchmarks33Exits The results are in. Overall, it was a great year in venture, even assome participants were left clapping on the sidelines. Last year, wesaw nearly $340B of venture capital (VC) investment, cash runwayincreased across all sectors, Series A tech deal activity began toimprove, and IPO and M&A activity hit the highest point since 2021.Barring a larger, macroeconomic recession or a deflation of the AIbubble, we expect growth to continue and metrics to improvethrough 2026. we expect this trend to flip in 2026, anticipating moderate dealgrowth driven by the AI platform shift and growing investorconfidence. If history is a guide, we must temper our short-term expectations forAI. Many investors we speak with who have been active participantssince the dot-com era say this is the largest platform shift in theircareer, while also acknowledging a potential bubble. In the threeyears since ChatGPT’s release, five AI companies have achievedvalues three times higher than all dot-com-era IPOs. There are clearsigns of overexuberance in AI: high burn multiples, low revenue peremployee and high valuation premiums for companies with lessrevenue than their peers. We are certain AI will generate long-lastinggains, but we are equally certain the path to value creation andinnovation is marked with expensive and perhaps necessarymistakes. The challenge is catching the AI wave, without gettingsucked under by the backwash of overhype. These multi-billion-dollardeals may be venture inname, but they havefundamentally differentrisk and return profilesthan traditional early-stage venture.” After 40 years on the job, some guys buy themselves a Corvette.Masayoshi Son bought a $30B stake in OpenAI. The SoftBank founderwasn’t the only VC breaking the mold in 2025. The definition ofventure has expanded to include a new class of mega funds andcompanies. The power law distribution in venture has a heavier headthan ever, with the top 1% of all venture deals accounting for one-third of all investment. These multi-billion-dollar deals may beventure in name, but they have fundamentally different risk andreturn profiles than traditional early-stage venture. The fundsinvested in these companies look more like private asset managerspaid with fees, not with carried interest. From Masayoshi Son’s $30B overtures to the quiet resurgence ofsub-$100M deals, venture is playing a more complex tune than everbefore. We enter 2026 mindful of the dot-com echoes and the heavy-headed power law, but we remain bullish. In periods of platformshifts, hype is unavoidable. But when the noise dissipates, trueinnovation remains and lays the foundation for the next decade oftechnological growth. At the other end of the spectrum, the funding environment is verydifferent. The bottom half of US VC-backed companies by valuationaccounted for just 7% of all investment. While capital invested insub-$100M deals picked up, deal activity trended down. However, Marc CadieuxPresidentSVB Commercial Bankmcadieux@svb.com Mark GallagherHead of Investor CoverageSVB Commercial Bankmgallagher@svb.com 5 AI companies have achieved 3x more valuethan all IPOs during the dot-com era Venture funds struggle to raise, hitting seven-year low; those that close take longer to do so VC investment hit second-highest year, one-third of capital went to the top 1% of companies Compounding challenges: Revenue required tofundraise is highertoday, yet growth is slower Jump to Page Jump to Page Median seedvaluationshave nearly doubledsince 2021 given high demandand low supply $4.4T locked in US VC-backed tech unicorns:secondary transactions ease liquidity thirst 2025 IPOs had a wide range of revenue,which many active unicorns surpassed Geopolitics and US policy push defensetech from covert to overt “The secondary market is not a substitute forIPOs.It’s a liquidity mechanism thatextends runway, relieves pressure ofcompanies going public too early, rewardsemployees, and dampens volatility alongthe exit path—but it doesn’t eliminate theneedtogopublic.At scale, once companiesreach tens of billions of dollars in value, therejust isn’t enough private capital to supportsustained growth and liquidity. The publicmarkets are still necessary, so IPOs may bedelayed for longer durations, but for mostcompanies they’re not optional, onlydeferred.” “Almost every LP of importance tells us thesame thing right now: We’re swimmingtoward early stage and away from the largeplatform funds.They understand thereturn dynamics, and they alsorecognize that big platforms have had ahard time truly operating at the earlystage.People are looking for the nextgeneration of managers, and they don’twant the overhead of big firms that haveeffectively turned into asset-managementbusinesses.” “I’ve seen several big tech waves over my ca