Alexandr Wang: “This is a Once-in-a-Civilization Opportunity”

Predictions and Expectations:

• Alexander Wang expects AI agents to help people get medical care to ultimately exist, noting that while the timing was wrong for his early venture, the concept is "very real."
• He believes that in the years following Scale's founding, data would become the critical bottleneck for training models, where one would "press a button" to access it.
• Wang expects that if AI model progress stopped today, there would still be "decades and decades" of total upheaval in the economy due to diffusion and adaptation.
• He predicts that the number of businesses on Meta's platforms will grow from 200 million to billions as an "explosion of entrepreneurship" occurs.
• Wang believes that the scarcity in the future will shift from intelligence and agency to "vision and ambition," requiring a clear view of how the world should look in "5 to 10 years."
• He expects that in a decade of hindsight, it will be obvious that "intelligence became abundant and that agency became abundant."
• Wang believes humanity will change more in the "next decade" than it has in the "past 100 years."
• He expects that "every wave [of AI] is 10 times bigger than the past wave," with new modalities and form factors driving dramatic growth.
• Alexander Wang is not sure if the decline in computer science majors will persist or if the skills needed to create good agents will become less relevant as abstraction layers rise.
• He believes that "the best AI products haven't even been developed yet."

Timelines and Milestones:

• Wang plans to launch a new "harness" "soon" to empower developers and agentic developers.
• He expects "bigger models" on the way within the Muse Spark line that will be "much more competitive with even the very best models that are out there today."
• He states that AI models will continue to become "more and more powerful" over the "next decade."
• Wang indicates that the transition of the world to fully embrace AI technology is happening today, though the world is currently "really just barely even ready."

Technology and Product Direction:

• Meta plans to continue shipping updates on the Muse Spark line and launching "open source models" to empower a "decentralized world of AI capability."
• The organization is focused on building a "harness" that is "extremely reliable," "very extensible," and capable of scaling to complex "multi-agent setups."
• Meta aims to allow models to plug into "all of the available harnesses" to empower "combinatorial innovation."
• The company believes in a future where billions of people have their own "personal superintelligence" tailored to their context and goals.
• Wang predicts that the next evolution will involve "agentic looping" and systems that can "spend like 1,000x more or 1 millionx more on tokens to drive an outcome in a continuous feedback loop."

Market and Industry Outlook:

• Wang predicts that startups leveraging AI agents can easily out-compete incumbents, shifting the dynamic from "David versus Goliath" to "Mecha-Goliath versus Goliath."
• He believes that the next decade will see "unprecedented opportunity for humans," including new sciences and solutions to health and biology problems "forever unsolved."
• Wang expects the global economy to undergo significant "upheaval and change" as the world adapts to existing AI technology, even if model improvements stall.
• He believes that "agency expansion" is a core driver, helping people accomplish things "they couldn't have ever dreamed of before."

Company Plans:

• Meta is building a Frontier Lab with a "zero-based build" approach, having already launched Muse Spark 1 and Muse Spark 1.1 within "nine months" of starting the effort.
• The company is working on a "harness" to help developers orchestrate agents and manage complex ecosystems.
• Meta plans to offer "a thousand dollars of free credits" for the Spark API to the audience.
• Alexander Wang believes the company must prepare "enterprises and governments" to adapt to new technology and figure out how to "secure the world from a biosecurity perspective or a cybersecurity perspective."

Financial Guidance:

• Alexander Wang notes that Muse Spark is "8X cheaper" than Opus, and converting to Opus dollars means the value is "a lot more."
• He mentions that the first wave of AI (self-driving cars) was "pales in comparison" to the second wave (LLMs), and the third (coding agents) was "probably 10 times bigger."

Risks and Caveats:

• Wang warns that if entrepreneurs go "too much with the herd," they will "get immensely confused and will end up nowhere."
• He identifies risks regarding "biosecurity" and "cybersecurity" that must be managed as the technology evolves.
• Alexander Wang states that the world is "really just barely even ready" for the full adoption of AI technology today.

Confidence and Disagreement:

• Alexander Wang expresses strong confidence that AI progress is on an "incredible exponential" curve and that the trend "will just continue."
• He disagrees with the focus on debating "when exactly" superintelligence will arrive, calling such debates "short-sighted" given the inevitability of powerful models.
• He believes with conviction that "systems thinking" and "rigorous thinking" are "never gonna go out of style" despite rising abstraction layers.
• Wang is not sure if the path for computer science majors has fundamentally changed or if the "move" for young people is shifting too far toward being "word cell and less shape rotator."