Amplify

Amplify Partners is a venture capital firm supporting technical founders, including engineers and AI researchers, through early investment and hands-on business support. Its AI investments span infrastructure, applications, and research commercialization: Amplify identifies itself as the first investor in Runway, Luma, and Modal, alongside data tools such as dbt, Dagster, Hightouch, and Hex. In June 2026, it announced participation in Engram’s funding, backing work on agent memory and continual learning.[3][2][7]

On June 23, 2026, Sarah Catanzaro announced Amplify’s participation in Engram’s $98 million funding. She connected the investment to two needs: AI products that adapt to users’ preferences and agents that acquire new skills from experience. The $98 million was Engram’s total announced funding; Amplify’s individual investment amount was unspecified.[7]

Catanzaro described Engram as initially focused on Cartridges, with expansion toward sparse memory fine-tuning. She credited Sabri Eyuboglu and a Stanford team with Cartridges research: a compact key-value (KV) cache trained offline through self-study on a document corpus, then loaded at inference. This approach would let a model reuse a prepared representation of documents or conversation logs rather than repeatedly placing the entire material into its prompt. She saw compact context representations and continuing memory updates as complementary directions for personalization and learning.[7]

The investment followed years of relationships with researchers. Catanzaro recalls meeting Eyuboglu through his Meerkat data-inspection library in 2021, Jessy Lin at an Amplify dinner in 2022, and Jack Morris in 2023 while he was a Cornell PhD student. She met Dan Biderman in 2025 and credits him with bringing the researchers together to commercialize their work.[7]

In her August 26, 2025 essay, “The AI research experimentation problem”, Catanzaro argued that generating ideas and implementing code would not, by themselves, substantially accelerate AI research. Her emphasis was on designing rigorous experiments, running them reliably, and analyzing their results. Combining several hypotheses in one training run can conserve compute, she explained, while making it harder to identify which intervention caused a performance change.[6]

Catanzaro also explained why apparently successful training can mislead: loss can continue falling despite bugs that degrade model quality. Reproducing a run with the same code, configuration, and data helps isolate failures. She distinguished three things an experiment implicitly tests—the research idea, its implementation, and its infrastructure—and cautioned that predictable loss curves do not guarantee predictable capability gains at larger scales. Her proposed improvements combined better evaluations, debugging, and observability with changes to organizational incentives and processes.[6]

These later AI investments and tooling arguments sit within an earlier organizational focus on technical company builders. Amplify’s founding premise covered distributed computing, developer tools, data, machine learning, and cybersecurity.[3]

Amplify was founded in October 2012. Founder Sunil Dhaliwal recalls spending fourteen years in venture capital before starting the firm, developing a belief that early-stage investing should rest on focus, domain expertise, and willingness to take risk.[4][5]

A consequential change came on October 21, 2020, when Amplify announced its first Select Fund. The $100 million fund targeted growth-stage investments in existing portfolio companies, alongside the $275 million Amplify Partners IV fund for initial Seed and Series A investments. Amplify said it deliberately sized the core fund to support companies throughout their lifecycle while preserving its early-stage focus; Select provided a separate means to make larger follow-on investments.[4]

Amplify retained that division in subsequent fund generations. On June 13, 2022, it announced a $400 million early-stage Fund V and a $300 million Select V fund for existing companies. On June 5, 2025, it announced $900 million across three funds: $400 million Fund 6 for first investments, $300 million Fund 6 Select for existing companies, and $200 million Amplify Bio, its first dedicated digital biology fund.[3][2]

Amplify Bio formalized a sector in which the firm had already invested. Amplify dates its Recursion investment to 2016 and names subsequent relationships with Chai Discovery, Axiom, Achira, Bexorg, and Tahoe Therapeutics. The June 2025 announcement also welcomed computational biologist and investor Elliot Hershberg as a Partner. The dedicated fund targeted founders combining biological research with programming and AI expertise.[2]

Amplify’s support extends beyond capital through Amplify Build, its hands-on platform team. In June 2022, the firm reported seven investment-team members and seven Build experts covering technical sales, recruiting, technical marketing, data, and design. Its June 2025 announcement described the Build Team as former sales, marketing, and recruiting leaders at technical companies, and reported a community of more than 150 technical founders exchanging practical advice on matters such as hiring engineers and usage-based billing.[3][2]

Amplify said in June 2025 that it had helped dozens of AI researchers commercialize their work, naming Datology, OctoML, Covariant, and CedarDB as examples. Its technical publishing also includes Catanzaro’s Projects to Know newsletter, which the firm dates to 2019 and describes as covering AI open-source projects and papers.[2]

In October 2020, Amplify reported 13 team members, more than 70 Seed and Series A investments since inception, and backed teams in more than 25 cities. It also reported that its active portfolio had raised more than $3 billion in investor capital—a portfolio-company financing measure—and cited the IPOs of Fastly and Datadog and acquisitions of BlueData, Buddybuild, Chef, and Conjur.[4]