Skip to content
Technology · North America

Cerebras CEO says local, purchase-capable personal AI assistants are close, Feldman tells TechCrunch Disrupt

Cerebras Systems CEO Andrew Feldman told TechCrunch Disrupt 2026 that personal AI assistants running locally on devices are close to reality. He said those assistants will "get to know you very well" and "will certainly have access to credit cards and be able to buy things for you."

AI-assisted synthesis · Based on 3 outlets' reporting · Full text read from 1 of 3 outletsHow it's made

3 outletsLatest reporting 3 min read
Coverage: Center OnlyL 0 · C 3 · R 0
Updat3 graphic: Cerebras Systems
Updat3 graphic

What happened

Key facts

  • Feldman said personal AI assistants that run locally on devices are not far away (per TechCrunch).
  • Feldman said those assistants will do some work on device and some work going back to the cloud (per TechCrunch).
  • Feldman said those assistants "will get to know you very well." (per TechCrunch).
  • Feldman said those assistants "will certainly have access to credit cards and be able to buy things for you." (per TechCrunch).

The brief

Andrew Feldman, chief executive of Cerebras Systems, told the TechCrunch Disrupt 2026 audience that personal AI assistants that run locally on consumer devices are close to becoming mainstream.

Speaking on stage at the conference, Feldman said the agents will split work between local hardware and cloud services, explaining "Just like your cell phone, they’ll do some work on device and some work going back to the cloud." He characterised the technology shift as driven by progress in consumer AI agents and chip design.

Feldman pointed to recent consumer-facing projects as examples of the trend, naming Meta’s Muse and the startup Instinct. TechCrunch reported that Instinct is valued at $10 billion, and Feldman used such advances to argue that AI is finding footing with everyday users.

He described how future assistants would form detailed profiles of users, saying they "will get to know you very well," and he added that people will place a security perimeter around those devices.

On the question of commerce, Feldman told the Disrupt audience that personal assistants will be capable of handling payments: "One of the reasons I love being in infrastructure is because I don’t need the consumer-facing idea … I like building the platforms that other people’s ideas take flight on," he said, and he explicitly said the assistants "will certainly have access to credit cards and be able to buy things for you." That claim frames forthcoming devices not only as information tools but as agents with purchasing capabilities tied to users' payment methods.

Feldman located Cerebras’s role in the ecosystem as an infrastructure and chipmaker, saying he prefers building the platforms that enable consumer innovation rather than consumer-facing products themselves. His remarks at Disrupt emphasised hardware and chip design improvements as enablers of the local-assistant vision, with the cloud still part of the computation mix.

TechCrunch’s account records these statements as made on stage at Disrupt; headlines from other outlets reflected the same central claim that Cerebras sees home AI assistants moving toward mainstream adoption and noted Feldman’s comments at the event. TechCrunch’s article includes direct quotations from Feldman and cites Instinct’s valuation as context for his remarks.

Feldman’s presentation frames a near-term pathway to assistants that combine on-device processing, cloud support, deep familiarity with users, and integrated payment access. The session underscored Cerebras’s positioning around the infrastructure that would support such consumer AI appliances, while attributing the consumer-facing ideas to other companies and startups.

Why it matters

  • Consumers who adopt these devices would face automated purchasing actions because Feldman said assistants "will certainly have access to credit cards and be able to buy things for you," creating a direct mechanism for financial transactions tied to AI.
  • Cerebras, as a chipmaker and infrastructure provider, stands to gain from demand for local, on-device AI compute, a market dynamic Feldman cited when he linked the vision to advances in consumer hardware and chip design.
  • The development trajectory Feldman described relies on a hybrid model of on-device processing plus cloud connectivity, meaning both device manufacturers and cloud service providers have documented roles in enabling these assistants.

What to watch next

  • Whether Cerebras or other chipmakers announce product details or partnerships to support on-device personal AI after Feldman’s TechCrunch Disrupt remarks.
  • Whether startups cited by Feldman, including Instinct, disclose further funding or product plans following the valuation noted by TechCrunch.
  • Whether consumer platforms like Meta publish technical or product updates for projects such as Muse that advance local assistant capabilities.

How it's being covered

Coverage analysisShow details
3 source articles · 3 unique outlets
Coverage mix: Center Only· 3 Center
Where the outlets are based: US/Canada 67% · Global 33%
Coverage mix
Center Only
Left
Center
Right
0 outlets
3 outlets
0 outlets
The marker shows coverage balance: each unique outlet counts once.
Geography
Where the outlets covering this story are based.
3 unique outletsDominant coverage: US/Canada
Coverage imbalance
  • Coverage is currently concentrated among Center-rated outlets (3 of 3 rated sources); comparatively few Left or Right-rated outlets have published on the event.
How we measure coverage

One source is from a wire service (AP, Reuters, AFP or Bloomberg), whose reporting other outlets often republish.

How outlets' accounts compare

Updat3 does not yet have enough full reports to compare how outlets framed this story.

How we got here

Immediate backdrop: public work showing large language models could run in constrained environments—Meta released the LLaMA family in February 2023 and open-source derivatives like Mistral (October 2023) enabled smaller-state models, while QLoRA (November 2023) demonstrated low-memory fine-tuning that industry groups cited as enabling on-device deployment.

Structural roots: the shift toward edge-capable AI built on the 2017 Transformer architecture ("Attention Is All You Need," Vaswani et al., 2017), the 2018 surge in pretraining exemplified by BERT (Devlin et al., 2018), and hardware efforts such as Cerebras Systems' founding in 2016 and its wafer-scale engine demonstrations by 2019 that aimed to accelerate large-model training.

Original sources

3 articles from 3 outlets

Every story links to the reporting it was built from. How we choose sources

Spotted a factual error, a broken link or a misleading summary? Tell us.