✌️ Heartcore Insights
Hi there,
Welcome to the 116th edition of Heartcore Insights, curated with 🖤 by the Heartcore Team.
If you missed the past newsletters, you can catch up here. Now, let’s dive in!
Does Gen AI Favour the Incumbent? - Henry Gladwyn
While Generative AI is new, human responses to technology are not. History can guide us, with the rise and fall of AOL Time Warner (“ATW”) offering a parallel to the current AI landscape.
AOL's trajectory should serve as a cautionary tale for companies like NVIDIA, highlighting how dominance in a new industry can be fleeting.
ATW was once a powerhouse, merging content and distribution to become one of the most valuable companies. However, today, AOL and Time-Warner are separate entities. Understanding their decline helps us grasp the erosion of incumbent advantage.
The ATW merger relied on AOL distributing Time-Warner’s content. The internet's explosion of information reduced the value of such content. Time-Warner thrived in a scarce information environment favouring broad appeal. But in the internet era, this content became less valuable compared to niche, freely distributed information.
Classic disruption theory applies: AOL’s paid services were upended by companies like Alphabet and Meta, which built business models around free, ad-supported content. ATW might have pivoted to ads, but this would have meant abandoning their lucrative portal business. When information was scarce, value lay with those who distributed it. With information abundant, value shifted to those who sourced it, like search engines and social networks.
Today's tech incumbents thrive on automation and prediction. AI, however, is creating an abundance of both, suggesting these incumbents might face a fall similar to ATW’s. Key AI inputs - data, computation, and distribution - should be examined through the ATW lens:
Data: Incumbents have massive data troves essential for AI training. Yet, as AI generates synthetic data, incumbent data may lose value, much like broad audience magazines did. The value may shift to new areas in the data chain, akin to how content value moved to platforms like Alphabet.
Computation: This includes top engineers and access to computing power. While engineers remain crucial, their decisive edge might diminish as skills diffuse and AI improves coding. Compute access, a current incumbent strength, could be challenged by new entrants offering cheaper, utility-based compute models.
Distribution: Traditionally seen as an incumbent advantage, it might lose relevance in an era of automation and prediction abundance. Future distribution could favour those catering to AI agents optimising for the best solutions.
Tech history shows that industries learn from past disruptions. Big Tech's proactive strategies, such as Zuckerberg's bold moves at Meta and Microsoft's early bets on OpenAI, highlight efforts to avoid being caught off guard. Even outside Big Tech, content owners' push into streaming is partly driven by a desire to avoid past mistakes.
All this underscores the importance of watching for structural changes, as incumbent business models may soon be ill-suited to a world of abundance.
AI’s $600B Question - David Cahn
In September 2023, an analysis by Sequoia called AI’s $200B Question examined the gap between revenue expectations from AI infrastructure investments and actual revenue growth in the AI ecosystem. This gap, described as a “$125B hole,” indicated a significant revenue shortfall.
Recently, Nvidia became the world’s most valuable company, prompting a reassessment of this analysis. The updated findings reveal that AI’s $200B question has grown into a $600B question. The calculation involves doubling Nvidia’s run-rate revenue forecast to account for total AI data centre costs, then doubling again to reflect a 50% gross margin for AI compute end-users.
Several changes since September 2023 include:
The GPU supply shortage peaked in late 2023 but has now eased. GPUs are more accessible, reducing previous procurement challenges.
Nvidia reported significant data centre revenue from large cloud providers, with Microsoft likely representing 22% of Q4 revenue. Hyper scale CapEx has reached historic levels, with Big Tech companies continuing to invest heavily in GPUs.
OpenAI’s revenue increased to $3.4B from $1.6B in late 2023. Despite some startups scaling revenues, OpenAI remains the clear leader. For AI companies to sustain consumer interest, they still must deliver significant value.
The previous analysis assumed major tech companies could generate substantial new AI revenue. Even with generous estimates, the $125B hole has now expanded to $500B.
Nvidia’s B100 chip offers 2.5x better performance for only 25% more cost. This is expected to drive another surge in demand, potentially leading to a supply shortage later this year.
Rebuttals to the previous analysis suggested that GPU CapEx is like building railroads, anticipating future demand. However, several key points challenge this view:
Physical infrastructure often has intrinsic value and monopolistic pricing power, unlike GPU data centres, which are becoming commoditised. New entrants in AI cloud computing increase competition, driving prices down.
Semiconductors, unlike physical infrastructure, improve rapidly. Nvidia’s ongoing production of next-gen chips like the B100 leads to faster depreciation of older models. This rapid improvement cycle diminishes the long-term value of current chips.
Winners vs. Losers: Infrastructure-building periods always have winners and losers. AI, as the next transformative technology, will create significant economic value. Declining GPU computing prices benefit long-term innovation and startups, primarily harming investors rather than company builders.
AI holds the potential to be a generation-defining technology wave. Companies like Nvidia play a crucial role in this transition and will remain important. While speculative frenzies are common in technology, maintaining a level-headed approach allows for the building of important companies. It’s crucial to avoid the delusion of quick riches from AGI and recognise that the journey will be long but worthwhile.
Beyond Benchmarks 2024, Emergence Capital
Robotics, FOMO, Scaling Laws & Technology Forecasting, Michael Dempsey
10 years after "Growth Hacking", Andrew Chen
What’s going on here, with this human?, Graham Duncan
Consumer retention, Bryan Kim
Define user engagement, Reforge
🇪🇺 Notable European early-stage rounds
Byway, a UK-based startup that is bringing journey-based flight-free travel to everyone, raises £5.04M with Heartcore 🖤/Eka Ventures - link
Soda, a Belgium-based data quality platform, raises $14M with Singular/Point Nine - link
bunch, a Germany-based tech platform that allows investors to syndicate deals across private markets, raises $15.5M with FinTech Collective/Cherry Ventures - link
Dust, a France-based provider of enterprise AI assistants, raises $16M with Sequoia Capital - link
Lakera, a Switzerland-based generative AI security startup, raises $20M with Atomico - link
Hive, a Germany-based operations platform that focuses on streamlining commerce operations, raises €28.2M with Earlybird/Tiger Global - link
Again, a Denmark-based startup that makes chemicals out of carbon dioxide, raises $43M with Google Ventures (GV)/HV Capital - link
🇺🇸 Notable US early-stage rounds
Substrate Labs, a startup that aims to simplify AI application development, raises $8M with Lightspeed - link
AuthZed, an authorization-as-a-service platform, raises $12M with General Catalyst - link
Phaidra, a virtual plant operator, raises $12M with Index Ventures - link
Caldera, a rollup-as-a-service platform, raises $15M with Founders Fund/Sequoia Capital - link
Sift, an observability platform for hardware sensor data, raises $17.5M with GV - link
Norm Ai, a regulatory AI platform, raises $27M with Coatue - link
Armada, an edge computing startup focused on satellite connectivity, raises $40M with Microsoft’s VC Fund (M12)/Founders Fund - link
Function, a membership service for health testing, raises $45M with a16z - link
🔭 Notable later stage rounds
Hebbia, a US-based AI-powered document search startup, raises $100M with a16z - link
Kandji, a US- based Apple endpoint management and security platform, raises $100M with General Catalyst - link
Chainguard, a US-based open source security startup, raises $140M with Redpoint/Lightspeed - link
Vanta, a US-based security and compliance platform, raises $150M with Sequoia - link
Saronic, a US-based company that designs and manufactures autonomous surface vessels for defence, raises $175M with a16z - link
HarmonyCares, a US-based provider of value-based in-home longitudinal care, raises $200M with General Catalyst - link
Astranis, a US-based company specializing in geostationary communications satellites, raises $200M with a16z - link
Formation Bio, a US-based AI-driven drug developer, raised $372M with a16z/Sequoia - link
Helsing, a Germany-based defence AI company raises €450M with General Catalyst/Elad Gil/Accel - link
🖤 Heartcore News
Congratulations to Byway on their latest funding round of £5M to bring journey-based flight-free travel to everyone, led by Heartcore 🖤 with Eka Ventures. 🚞
Congratulations to our portfolio company GOURMEY on being the very first cultivated meat company ever to file for approval in the EU. 🥩
Our portfolio company Issuu has been acquired by Bending Spoons, a content powerhouse based in Italy. ⏭️
Our Web3 team reconnected with our portfolio founders from Hylé, Rhinestone, Superform Labs, Aave, dYdX, and many others, and saw them go on stages across Belgium at EthCC. ✌️
Listen to Neil Chatterjee, CEO, Andrena, a Heartcore Web3 🖤 portfolio company, on Logan Jastremski’s podcast. 🎙️
Shoutout to the new faces at Heartcore 🖤, Frederik & Bérenger! We will announce a few more in the coming weeks. 👏
Thank you for being a loyal subscriber of our monthly Insights - please feel free to share this newsletter with anyone you’d think would appreciate it!