What is the main shift Tony Kim observed in AI investing this year compared to prior years?

The focus has expanded from model development to the physical infrastructure supporting AI: data center design, chips, power, cloud infrastructure, and now space-based compute. Additionally, AI is moving into physical embodiment through autonomous vehicles, humanoid robots, and low Earth orbit satellites used for data and computation.

When does Tony Kim expect to see real adoption of autonomous vehicles and robots?

Within the next five years, with cars already seeing adoption and self-driving trucks expected to begin operating in Texas without drivers at the beginning of next year. Kim noted that a million-plus cars and humanoids in production could be possible within that timeframe.

How could space-based satellites address AI infrastructure challenges?

Low Earth orbit satellite constellations can serve two purposes: they can provide satellite imagery as a new data modality for AI models, and they can host compute clusters that beam inference tokens to Earth, potentially bypassing terrestrial constraints like energy supply, permitting, and regulations that burden conventional data centers.

RoboticsAutonomous DrivingAI Stocks & MarketsTop Companies' AI MovesTop Companies AIAug 17, 2026Aug 18, 2026, 06:32 JST

AI moves beyond software into space, robotics, energy infrastructure

Send on LINE

3 Key Points

What happened
BlackRock's head of global technology, Tony Kim, returned from his 13th annual tech tour across San Francisco and Silicon Valley in June with 35 colleagues, observing that AI investment is expanding beyond model development into data center design, physical systems like autonomous vehicles and humanoid robots, and low Earth orbit satellites. The conversation has shifted from pure AI models to compute-centric infrastructure and the supply chains supporting it.
Why it matters
As AI infrastructure demand grows, scarcity is emerging in energy, materials, and semiconductors—challenges that space-based compute may partly address. The value accrual in technology markets is now concentrated in compute and model-centric companies rather than broader software players, signaling a reshaping of which technology businesses will benefit most. Companies tied to the physical buildout of AI (power, chips, data centers, satellites) may be positioned differently from service-oriented software vendors.
What to watch
Real adoption of physical AI systems is expected within the next five years, with Kim noting that a million-plus cars and humanoids in production could be possible. Self-driving trucks without drivers are expected to begin operating in Texas in the beginning of next year, marking a shift from pilots to operational deployment. The integration of satellite constellations for both AI data (via satellite imagery) and compute infrastructure represents a new frontier in AI's infrastructure layer.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

Context & Analysis

The tech landscape is undergoing a fundamental reorganization driven by AI's maturation from software innovation to physical infrastructure buildout. Where AI investment conversations two years ago centered on language models and software platforms, the discussion has now moved to the hardware and energy foundations required to power those systems at scale. This shift reflects a recognition that the bottleneck to AI growth is no longer algorithmic but logistical: energy supply, semiconductor capacity, data center construction, and the power and cooling infrastructure supporting them. Kim's observation that value is concentrating in compute and model-centric technology companies signals a repricing of the market, with winners being those tied to the physical layer (chip makers, power companies, data center operators, and now satellite providers) rather than pure software or application-layer businesses. The emergence of space-based compute as a serious topic underscores how scarcity—particularly in energy and terrestrial real estate for data centers—is driving companies to explore unconventional solutions. The timeline Kim outlined, expecting mass adoption of autonomous vehicles and humanoids within five years, suggests that the next phase of AI's economic impact will manifest not in better chatbots but in the physical economy: transportation, manufacturing, and logistics. This represents both an investment opportunity and a structural risk for technology companies whose business models were built around software economics rather than hardware and infrastructure buildout.

FAQ

What is the main shift Tony Kim observed in AI investing this year compared to prior years?: The focus has expanded from model development to the physical infrastructure supporting AI: data center design, chips, power, cloud infrastructure, and now space-based compute. Additionally, AI is moving into physical embodiment through autonomous vehicles, humanoid robots, and low Earth orbit satellites used for data and computation.
When does Tony Kim expect to see real adoption of autonomous vehicles and robots?: Within the next five years, with cars already seeing adoption and self-driving trucks expected to begin operating in Texas without drivers at the beginning of next year. Kim noted that a million-plus cars and humanoids in production could be possible within that timeframe.
How could space-based satellites address AI infrastructure challenges?: Low Earth orbit satellite constellations can serve two purposes: they can provide satellite imagery as a new data modality for AI models, and they can host compute clusters that beam inference tokens to Earth, potentially bypassing terrestrial constraints like energy supply, permitting, and regulations that burden conventional data centers.

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.