
Misha Laskin’s strategy for competing in AI
All key points
Read the points below. Tap a timestamp to play it in the video above.
- 01
Laskin says Reflection will build high-quality open models, deploy them to enterprise, public-sector, and sovereign customers, and build a generational company around them. (55: 06.661)
Play this point here · - 02
He wants Reflection to become a standalone leader and help create a Western open-model ecosystem comparable in quality to China’s. (55: 36.061)
Play this point here · - 03
He calls Chinese models “a gift to the world” and says America needs the best open-model stack, including models and chips, so others choose American technology. (47: 59.948)
Play this point here · - 04
He says business advantage comes from developing the supporting technology stack and earning trust with enterprise, public-sector, and sovereign customers. (37: 03.061)
Play this point here ·
AI summary
What is Reflection’s strategy? According to Misha Laskin, Reflection plans to keep building high-quality open models, deploy them to enterprise, public-sector, and sovereign customers, and build a generational company around them. He also wants Reflection to become a standalone leader while contributing to a broader Western ecosystem of open-model companies comparable in quality to China’s ecosystem.
(Evidence 0, 55:06.661; Evidence 1, 55:36.061) Why does he call open-source models a “gift”? Laskin describes Chinese models as “a gift to the world.” He argues that the United States should develop the best open-model stack, including models and chips, so that countries and allies choose to build on American technology rather than become dependent on Chinese infrastructure. (Evidence 2, 47:59.948) How would Reflection build a sustainable business?
Laskin says a Western open-model company can gain an advantage by developing the supporting technology stack ahead of competitors and earning trust with enterprises, public-sector organizations, and sovereign entities. The supplied passages do not specify a more detailed tactic aimed specifically at closed-source Chinese models. (Evidence 3, 37:03.061)
TalkOnPoint



