Headlines this week - Aug 3, 2025
A look at how capital is being deployed across future opportunities
This week in the future:
1 - China showcases its ambitious vision to lead the world in AI
China is promoting a vision of “global AI cooperation”. At a major conference in Shanghai, China presented its AI vision as a collaborative and open alternative to the US "America First" plan. According to the FT, this involved promoting open-source models and announcing new international organizations for AI governance.
The underlying goal is to build a self-sufficient AI stack. Behind the collaborative rhetoric, China is "girding for an AI battle with the U.S." by aiming for a fully independent AI supply chain. The WSJ reports this involves massive state investment in domestic chips, models, and manufacturing to insulate itself from American technology restrictions.
These efforts are already producing powerful new AI models. The strategy has started to bear fruit, with Chinese startups regularly releasing advanced models. This week, Chinese startup Z.ai launched a new open-source model family that shows competitive performance against leading Western proprietary models, including features like PowerPoint creation.
This intense rivalry could create dangerous geopolitical risks. The escalating US-China competition in AI is a "dangerous game," according to an FT op-ed this week. The piece warns that a scenario of "technological surprise" from either side (where one nation achieves an unexpected breakthrough) could dramatically escalate the risk of real-world conflict.
2 - Wall Street rewards Big Tech's massive AI spending, but questions remain
Strong results from Meta and Microsoft earned a market "blessing”. Both Meta and Microsoft saw their shares jump after strong quarterly results, which investors took as validation of their massive AI spending plans. A WSJ analysis called this reaction Wall Street's "blessing" for a combined $400bn AI spending spree.
However, Amazon was penalized for slower cloud growth. In contrast, Amazon's stock fell as its high AI spending was paired with underwhelming third-quarter projections. The FT notes that its cloud division's growth was seen as "disappointing" compared to stronger performance from rivals Microsoft and Google.
Meanwhile, skeptics warn that the AI investment race is a risky "Field of Dreams" strategy. Some analysts caution that the market's enthusiasm is premature. Richard Waters at the FT argues Meta’s current strong results are not from generative AI, making the massive new investments a "shot in the dark" based on a "Field of Dreams" fallacy: “building it” doesn't guarantee the right customers “will come”.
3 - New thinking is emerging on how AI will reshape the enterprise (with no humans in the loop)
Autonomous agents may be more effective in “digital transformation” than hybrid human-AI teams. In his "One Useful Thing" newsletter, Ethan Mollick applies AI's "Bitter Lesson" (the idea that general computation often beats human-designed knowledge) to how organizations work. He suggests that since companies’ processes are often chaotic, autonomous agents may find more efficient solutions than carefully planned human-in-the-loop systems.
4 - AI's disruption of the job market is accelerating and creating deep divides
CEOs are now proudly announcing workforce reductions. A WSJ article this week notes a new trend where CEOs are not just cutting jobs but seem "prouder" to do so and communicate the plans to investors. These announcements are often framed as a positive signal of tech-driven efficiency, with AI providing a new justification for sustainable leaner workforces.
The impact is hitting recent college graduates the hardest. The job market for new graduates is "already fragile" and AI is "wrecking" it further, according to the WSJ. As we’ve already discussed here, the automation of entry-level white-collar tasks is shrinking the number of available roles, making it significantly harder for this group to start their careers.
A massive pay gap is emerging between AI "superstars" and other professionals. The war for elite AI talent is creating deep inequality within the tech industry itself. The NY Times reports that top researchers are now negotiating pay packages of up to $250m, similar to NBA stars, creating a vast compensation gap.
The talent war continues, with Meta poaching again from Apple... The battle for top talent remains fierce, with Meta hiring its fourth AI researcher from Apple in just one month. Bloomberg reports that these departures have thrown Apple's foundational models team "into flux," demonstrating the direct competitive damage these hiring campaigns can inflict.
...but even Meta's massive offers can be rejected. Meta's aggressive tactics don't always succeed, as detailed in a WIRED article this week. The company reportedly approached over a dozen researchers at Mira Murati's startup with massive offers (one exceeding $1bn), but so far, none have accepted the roles.
5 - AI creates new security threats, fueling the cybersecurity boom
The hacking of an AI coding tool at Amazon reveals new vulnerabilities. A recent hack of an Amazon AI coding assistant has exposed a major new security risk. Parmy Olson at Bloomberg explains that the hacker used plain language prompts to trick the AI, revealing a "gaping hole" in security practices and adding a social engineering dimension to AI-related threats.
Cybersecurity incumbents’ need of new skills is fueling M&A deals. To counter these new threats, major cybersecurity firms are making large acquisitions for new capabilities. As an example, Palo Alto Networks is buying Israeli identity security firm CyberArk for $25bn, to protect against breaches by both human and AI "agents" accessing sensitive data.
The Palo Alto Networks’ deal is part of an "AI mania" hitting security software M&A. This acquisition is part of a larger trend of consolidation in the security sector, driven by AI. An FT Lex column compares the deal to Google's recent $32bn purchase of Wiz, another Israeli firm, arguing that security has become the "hottest niche" as more AI creates more data to protect from sophisticated threats.
6 - More signs of accelerating innovation (plus scientific and ethical hurdles) in genomics
The emerging embryo selection industry is becoming fiercely competitive. As the field heats up, new startups are entering the market with a competitive edge. A just-released white paper from new entrant Herasight directly critiques the claims of a rival (Nucleus, recently mentioned in our summaries), arguing its marketing has outpaced its science, signaling a new, more aggressive phase in the industry.
Trait prediction, not sequencing, remains the main technical and ethical challenge. The key hurdle is not sequencing an embryo's genome, but accurately predicting traits from that data. A deep dive on embryo selection by Scott Alexander this week notes that predicting complex traits is a moral gray area, and that the difficulty in validating these predictions creates a high potential for "fraud or mistakes".
Better methods and data are needed to unlock AI's full potential in biology. For AI to truly revolutionize biology, more than just “big data” is required. An essay in "Bioptimist" argues that current AI approaches are a vast oversimplification, like "using constellations for interstellar travel," and that better tools and data are needed to generate and interpret the right kind of biological data.
7 - Solar power's historic growth surge is set to continue
Solar generation is growing faster than any power source in history. The rise of solar power generation is the fastest in the history of electricity, according to a Goldman Sachs report featured in the FT. This "structural solar surge" is driven by a virtuous cycle of falling costs, zero marginal fuel costs, and the technology's modular nature.
New solar cell technologies could supercharge this growth. This rapid expansion is expected to be reinforced by ongoing innovation. An FT Rethink video report suggests that the latest breakthroughs in solar cell technology are poised to "supercharge" this trend, further improving the efficiency and economics of solar power for the future.
8 - Silicon Valley meets the physical world: AI opens the door to general-purpose robots
Progress is accelerating across all levels of robotic autonomy. A new framework from SemiAnalysis argues that recent AI foundational models have unlocked a critical level of "agency" for robots. This leap is described as the "genesis" for true general-purpose robotics that can navigate the complex, open world and eventually replace human labor.
Big Tech is investing heavily in this new paradigm. Major tech companies are actively pursuing this vision, as highlighted by a Reuters report on Amazon-backed Skild AI. The subsidiary has just unveiled a new general-purpose AI model designed specifically to power the next generation of multi-purpose robots.
All this happens in the context of a broader Silicon Valley shift toward the physical world. According to Christopher Mims at the WSJ, this trend could be seen as part of a broader pivot for Silicon Valley to "Move Slow and Build Things". This new strategy signals a strategic shift from purely digital products to tackling complex, physical-world problems like general-purpose robotics.
9 - The US U-turn on AI chip sales to China sparks intense debate
The White House has frozen chip export controls. To avoid jeopardizing trade deal negotiations with Beijing, the Trump administration has frozen restrictions on technology exports to China. The FT reports this includes a reversal of the recent ban on Nvidia's H20 AI chips, a move made to help secure a trade deal.
The move has created a sharp controversy over its strategic impact. The decision is highly controversial. The Economist notes that the lack of these chips was the key "bottleneck" limiting China's AI industry. In contrast, we’ve already discussed in previous weeks the narrative, fostered by Nvidia’s J Huang among others, claiming that bans, if maintained, would drive the creation of a fully, end-to-end Chinese stack, with Huawei emerging as a big rival to Nvidia
10 - The future of connectivity is moving beyond satellites… and fiber…
Stratospheric platforms are emerging as a new connectivity layer (alternative to satellites). Companies are now developing High Altitude Platform Stations—airships or drones that act as floating cell towers in the stratosphere. As described in IEEE Spectrum this week, this technology is being positioned as a direct competitor to satellite constellations for providing wide-area internet coverage.
On the ground, laser-based internet could help bridge the "middle mile”. Meanwhile, new technologies are also improving terrestrial connections. An IEEE Spectrum report details how Google spin-off Taara is using over-the-air lasers to beam high-speed internet across difficult terrain, solving the "middle-mile" problem where laying fiber optic cable is impractical.
LINKS:
1 - Population & natural resources
Biotech
Genomics
More signs of accelerating innovation (plus scientific & ethical hurdles)
Fertility
Falling fertility rates as a new existential risk for humanity
Water
Investments growing in desalination tech. But still lots of work to do
Desalination: the costly solution for a thirsty world
Materials
A new quantum phase of matter potentially useful for low-power computing
UC Irvine Unveils First Spin‑Triplet Excitonic Insulator in Quantum Matter
Ultra-realistic “mixed reality” displays could be on the way
A leap toward lighter, sleeker mixed reality displays
Energy
European and US utilities are raising capital to upgrade the grids
European electricity grids turn money into power
AI is making energy prices rise, also for consumers. Conflicts ahead…
Who Pays? AI Boom Sparks Fight Over Soaring Power Costs
Nuclear
China is using its massive scale to reduce construction costs of nuclear
China reins in the spiralling construction costs of nuclear power — what can other countries learn?
Renewables
Solar generation is growing faster than any power source in history
Innovation could even accelerate the trend
Can new technologies supercharge solar power? | FT Rethink
Geothermal energy is starting to be viewed as one more alternative
Superhot geothermal energy could unearth big power boost for the AI era
2 - Efficiency & Productivity
New Transport Technologies
Autonomous Cars
Waymo will partner with Avis (not Uber) in its Dallas’ robotaxi operation
Waymo taps Avis to manage robotaxi fleet in Dallas
Computing Infrastructure
Data Centers
Wall Street rewards Big Tech's massive AI spending
Meta shares jump on strong results as Zuckerberg sets out ‘superintelligence’ goals
Microsoft profits soar almost 25% on boom in cloud computing
Big Tech’s $400 Billion AI Spending Spree Just Got Wall Street’s Blessing
Amazon was an exception, penalized by weak AWS revenues
Amazon Shares Fall Because Cloud Unit’s Growth Wasn’t Enough for Wall Street
However, there are still plenty of uncertainties on AI investment returns
Communications
The future of connectivity is moving beyond satellites… and fiber…
Chips
Trump froze AI chip export controls to get a better trade deal with China
Donald Trump freezes export controls to secure trade deal with China
The move is already impacting Nvidia’s sales
Nvidia orders 300,000 H20 chips from TSMC due to robust China demand, Reuters reports
Arm could soon become a competitor of Nvidia
Arm to explore designing its own chips, CEO says
Samsung signed a massive foundry deal with Tesla, but will it be enough?
Tesla’s $16.5bn contract won’t be enough to drive a revival at Samsung
Quantum Computing
India wants to be a global hub for quantum computing technologies
Narendra Modi’s kingmaker aims to build Indian ‘quantum valley’
Meanwhile, Chicago could turn into a US national hub for the same thing
Why the Billionaire Pritzkers Got Obsessed With Quantum
Artificial Intelligence
AI: Apps, Agents
Agents
Fully autonomous agents may be more effective in the enterprise than hybrid human-AI teams
The Bitter Lesson versus The Garbage Can
B2C
Google keeps improving its search value proposition with AI
Google's AI Mode gets new 'Canvas' feature, real-time help with Search Live, and more
B2B
Amazon is investing in an AI app that generates TV shows, according to users’ prompts
Physical AI: Robots / Drones
AI opens the door to general-purpose robots
AI: Foundational Models
AI talent wars continue
Apple Loses Fourth AI Researcher in a Month to Meta’s Superintelligence Team
A.I. Researchers Are Negotiating $250 Million Pay Packages. Just Like NBA Stars.
Anthropic may now be worth $170bn
Iconiq set to lead $5bn funding round for AI start-up Anthropic
And they are looking for better monetization schemes
Anthropic unveils new rate limits to curb Claude Code power users
China showcases its ambitious vision to lead the world in AI
Chinese efforts are already producing powerful new AI models
Chinese startup Z.ai launches powerful open source GLM-4.5 model family with PowerPoint creation
AI: Security & Safety
AI safety
Intense US-China rivalry in tech could create dangerous geopolitical risks
Global AI rivalry is a dangerous game
AI’s negative side effects
Outsourcing mental tasks to AI models can cause human skills to atrophy
The risk of letting AI do your thinking
Cybersecurity
The hacking of an AI coding tool at Amazon reveals new vulnerabilities
Amazon’s AI Coding Revealed a Dirty Little Secret
Cybersecurity incumbents’ need of new skills is fueling M&A deals. E.g. Palo Alto Networks
Intelligence Augmentation
Augmented Reality
Alibaba will launch its own smart glasses (similar to Meta’s Ray-Bans)
Alibaba to launch AI-powered glasses creating a Chinese rival to Meta
3 - Economic / Business trends
Tech & Geopolitics
Sharp controversy on the strategic impact of removing chip export bans
How Trump’s U-turn on chips could unleash Chinese AI
Emerging economic models
Jobs
AI's job market disruption accelerates. CEOs announcements proliferate
CEOs Are Shrinking Their Workforces—and They Couldn’t Be Prouder
The impact is hitting recent college graduates the hardest
AI Is Wrecking an Already Fragile Job Market for College Graduates