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。

OpenAI拟在业务扩张加速之际将员工规模扩大一倍

这家估值7300亿美元的初创公司计划在2026年底前将员工扩充至8,000人,力求缩小与竞争对手Anthropic的差距。

多项研究显示数字设备正在让我们变笨

一些学者开始重新拿起纸笔,但并非总能彻底杜绝手机和笔记本电脑。

我们如何区分好的和坏的AI?

来自新技术崎岖前沿的更多启示。

芯片测试的考验与软银集团的失足

但随着人工智能的应用不断演进、全球对更强大芯片的需求持续增长,半导体产业从芯片测试到存储芯片生产的整个供应链都在努力加速以跟上步伐。

人工智能能帮我找房吗?

新科技正在改变找房方式——但机器视觉和高斯点云会取代房产门户网站与房产中介吗?

普华永道美国负责人称,抗拒AI的合伙人不适合留在公司

在科技削弱其业务之际,这家咨询公司着手全面改革定价与服务。
设置字号×
最小
较小
默认
较大
最大
分享×