Chile and ElevenLabs: 44M in AI voice credits, but the company picks the winners

By El Ecosistema Startup · November 11, 2026.

Chile's Ministry of Science, Technology, Knowledge and Innovation and ElevenLabs have announced a memorandum of understanding to expand Chilean startups' access to voice and audio AI. The core commitment is up to 44 million platform credits over 12 months, to be distributed among eligible startups through a competition. According to the outlet's reporting, Minister Ximena Lincolao presented it as a bet on startups "that can compete globally."

Four tools are included: voice generation (text-to-speech and voice cloning), audio transcription, music generation and voice agents for conversations. Those selected will also receive support and technical assistance from the ElevenLabs team, subject to approval.

The division of roles is the most striking part. According to the reports, ElevenLabs sets the eligibility criteria, selects the beneficiaries and runs the program. The Ministry acts as a facilitator: it promotes the competition and connects with the local innovation ecosystem. In other words, the government does not decide who receives the credits.

There are also limits the announcement leaves unresolved. The text says the company "expresses its intention" to provide the credits: it is a memorandum, not a contract with guaranteed delivery. Nor has there been any information on how many startups will be selected, the dollar value of the credits or when the competition will open.

The company's background helps explain the official interest. ElevenLabs, founded in 2022, headquartered in London and led by Mati Staniszewski, focuses on voice AI. According to Diario Financiero, it is valued at $11 billion and counts investors such as Sequoia. Variety, for its part, reported that in September it completed a $300 million employee share tender offer that valued it at $22 billion, double its February valuation.

**Our take.** This deal is good news for anyone building voice products in Chile, but it should not be mistaken for industrial policy. The credits are an in-kind subsidy, and the party granting them is a vendor that, by handing them out, wins over future paying customers. It is a legitimate commercial move and, at the same time, a way of creating dependency: a startup that builds its core on a single API inherits its pricing, its policies and its roadmap.

The underlying risk ties into a trend we have been tracking: the companies that control AI distribution increasingly set rules that used to be public—in this case, who gets access to the technology. Having the government delegate the selection is practical, but it gives up judgment and transparency. It is worth demanding that the criteria and the results be made public.

There is a safety dimension the announcement does not mention. Voice cloning is one of the package's core tools, and it is also the raw material for audio deepfakes and voice scams. We are not saying the program implies such use, but democratizing the tool requires pairing it with consent verification and traceability. Trust is shifting from "it sounds like that person" to provable provenance.

In the short term, the sensible course for a startup is to narrow down a use case, calculate inference costs at list price for when the credits run out, and keep a plan B with another provider or with open models. In the long term, voice as the primary interface could bring services that are expensive today, such as personalized customer service, education or healthcare support, to many more people. That horizon of abundance will only become real if the local ecosystem ends up with products on the market, not just credits handed out.