Big bets on AI point to venture capital industry’s shift - FT中文网
登录×
电子邮件/用户名
密码
记住我
请输入邮箱和密码进行绑定操作:
请输入手机号码,通过短信验证(目前仅支持中国大陆地区的手机号):
请您阅读我们的用户注册协议隐私权保护政策,点击下方按钮即视为您接受。
观点 风险投资

Big bets on AI point to venture capital industry’s shift

The strategy of focusing money on larger deals comes at time of extreme indigestion for start-up investors

Depending on where you sit in the investment world, the venture capital business is either in rude health or facing something of an existential crisis.

Like many people in tech these days, start-up investors have backed artificial intelligence to the hilt. The latest evidence came with this week’s news that Databricks, a provider of software to gather and analyse large volumes of data, has raised another $10bn, one of the largest private investment rounds ever.

Their willingness to put up the kind of large amounts that would once have required Wall Street involvement shows how some of the biggest venture investors are navigating the AI boom with a distinct swagger.

But doubling down on AI has coincided with a period of severe indigestion for the world of start-up investing at large. The industry has barely begun to work its way through an immense overhang of investments from venture’s Zirp era — the period, ending in 2021, when a zero-interest rate policy brought a flood of capital into tech start-ups.

This has left about $2.5tn trapped in private unicorns, or companies with a valuation of $1bn or more. At least, that’s the combined value these companies claimed after their last fundraisings, according to PitchBook. When it comes to actually trying to cash in these chips through initial public offerings or the M&A market, the returns are likely to be a lot less. How much of the venture business will be left standing after the eventual reckoning is hard to tell.

First, consider the scale of the bet on AI. Databricks set out to raise $3bn-$4bn in its latest round, but chief executive Ali Ghodsi said that investors had offered $19bn (he decided to roughly split the difference).

Given the overwhelming level of demand, Databricks’ latest valuation doesn’t look outlandish. At $52bn before the addition of the new cash, it was up from $43bn 15 months before and roughly equivalent to 17 times its annualised revenue run-rate — hardly outrageous for a business growing at 60 per cent a year.

Private financing rounds of $1bn or more were once a rarity. It took the huge ambition of SoftBank’s Vision Fund and a handful of specialist late-stage investment groups to break the mould. Now, investors like Thrive Capital, which led the Databricks round, pride themselves on putting up $1bn single-handed.

Over the past two years, AI model builders OpenAI, Anthropic and Elon Musk’s xAI have raised nearly $40bn between them. Other sizeable investment rounds this week alone included $500mn for Perplexity, an AI-powered search engine, and $333mn for Vultr, part of a new band of companies running specialised cloud data centres to support AI. 

What makes this boom in private backing for AI all the more remarkable is that it comes against the backdrop of a broader collapse in venture investing. Compared with the boom year of 2021, before the interest rate cycle turned, the amount of venture capital invested two years later had plummeted by 55 per cent, to $161bn, according to PitchBook. In the first nine months of this year, fewer than half as many investors completed deals as in all of 2021.

Fewer, bigger funds pumping ever-larger amounts into an increasingly narrow range of companies, almost all of them in AI: it’s a long way from the model on which venture was founded, of spreading small amounts of investment seed corn widely in the hopes that the occasional big hit would make up for many misses.

But VC’s concept of itself has changed. In many ways, private capital markets for tech now rival Wall Street. Rates of return will necessarily fall as much larger amounts of capital are put to work in more mature companies, though the successful investors will no doubt point out that they stand to make better returns than similar-sized funds investing in other asset classes.

For many other venture investors, the situation has become little short of critical. After a brief boom in 2021, IPOs and sales to strategic buyers have fallen off a cliff. With less cash being returned, many of the investors which back VC funds are unwilling to put up more. Many start-ups that achieved unicorn status during the boom would rather cut costs and conserve cash than return to raise more money at a lower valuation. It will take time for this to work through the system, but the reality — that many Zirp valuations are no longer supportable — will be unavoidable.

Investors in the latest round of giant AI fundings will be hoping to escape a similar fate. Companies such as Databricks, which says it will turn cash flow positive this quarter, already look ready for an IPO. That could make 2025 a pivotal year for VC’s latest investment fad.

richard.waters@ft.com

版权声明:本文版权归FT中文网所有,未经允许任何单位或个人不得转载,复制或以任何其他方式使用本文全部或部分,侵权必究。

科技巨头为什么对“通用人工智能”众说纷纭

通用人工智能被誉为硅谷下一个重大突破,但它究竟是一个科学目标,还是一个营销流行语?

洛克希德•马丁向英国推销导弹防御系统

美国防务集团希望在地缘政治紧张局势加剧以及美国投资“金穹”之际,为英国建造一个新的导弹防御系统提供帮助。

军事力量逐步就位,特朗普接近对伊朗发动打击

美国总统暗示将在数日内采取行动,美国已准备好能够打击福尔道地下核设施的部队。

普京召开的投资论坛未能吸引西方公司

俄罗斯的盟友们也只是向圣彼得堡派遣了低级别的官员和商人,但印尼总统是个例外。

微软准备退出与OpenAI的关键性谈判

ChatGPT开发商计划转型为营利性公司,促使这家软件巨头制定应对预案。

FT社评:特朗普需要慎重考虑伊朗问题上的命运抉择

美国总统可能会被拖入中东另一场寻求促成政权更迭的愚蠢行动中。
设置字号×
最小
较小
默认
较大
最大
分享×