For the private markets, this means thinking about more AI, all the time. That said, I do think next year the rubber is going to meet the road for AI startups and giants alike. High compute costs, compressed margins, and soaring valuations and expectations will inevitably collide with reality. And for some, this will mean even more acquisitions and more acquihires than perhaps we’ve seen so far in the AI boom. 

I started asking around: Which startups would make smart acquisition targets for a tech giant in 2026?

“To unlock ‘real world’ AI like robotics, autonomous vehicles, smart factories, spatial computing, and embodied AI, tech giants need models that can reason about the real world in real time,” said Aidan Madigan-Curtis, Eclipse Ventures partner, via text. “Startups like Wayve, Physical Intelligence, WorldLabs, Bedrock Robotics, The Bot Company and GenesisAI, are already building simulation engines, sensor fusion stacks, and world models that learn from physical interaction—capabilities that would take incumbents years to replicate internally.” (Eclipse is an investor in Wayve.)

Madigan-Curtis gets at an essential question: In AI, when does it make more sense to acquire rather than build? Shensi Ding, CEO and cofounder at AI integration infrastructure startup Merge, points out an unconventional idea around finance (a widely touted AI use case): “Large AI players should acquire boutique investment banks and use historical financial models to train them. This work is highly specialized and requires domain expertise to really break through and build trust.”

Meanwhile, Morgan Blumberg, M13 principal, thinks that large foundation model companies will look to gobble up application layer companies with proven product-market fit. The obvious targets: coding tools, one of enterprise AI’s great 2025 success stories.

“In 2025, we saw Windsurf in the coding space attract strong interest,” said Blumberg via text. “While some like Cursor might choose to stay independent, I predict there will be attractive prices for assets like Factory, Codegen, Wrap, and others.”

Zach Lloyd, CEO and founder of agentic coding startup Warp, reinforced that developers are a key customer base: “AI giants should acquire an observability platform like Datadog or Sentry,” he said via email. “These tools sit where code meets reality (logs, errors, traces, and production failures) which is exactly the context AI needs to be genuinely useful to developers.”

This push to get enterprise right transcends foundation-model mainstays like OpenAI or Anthropic, and for some large companies, it might make perfect sense to buy a unicorn outright, said Jake Stauch, CEO and founder of Serval, which builds AI agents for IT. “They could look to acquire enterprise AI solutions in customer support or enterprise search, such as Sierra or Glean respectively,” he said. 

It’s worth saying: Pretty much any deals of this ilk coming to pass would be, well, a big deal. That said, any potential deal target deserves serious scrutiny. So much capital has flowed into so many of these AI businesses. And last time I checked, even in the most abundant situations, there are inevitably a finite number of generational public companies. 

This is the last Term Sheet of 2025, and when we’re back on January 5, it’ll be with our much-loved Crystal Ball prediction series. So, I’ll leave you with one prediction of my own: Next year, we’ll enter a period where the haze of flowing capital and buzzy rhetoric will clear just a little, and we’ll start to see who can actually go the distance. 

See you in 2026, 

Allie Garfinkle
X: @agarfinks
Email: alexandra.garfinkle@fortune.com
Submit a deal for the Term Sheet newsletter here.

The deals section will return in 2026. Subscribe here.