FTFL - Return on Innovation

Often regarded as the most innovative country in the world, does the United States have an ROI problem when it comes to innovation? Consider some comparisons:

• DeepSeek – founded just 1.5 years ago – reportedly spent only $6 million to develop R1, compared to OpenAI spending over $100 million to develop o1 
• BYD built a battery life 4x that of Tesla, while selling at an average price of $16,700 vs. Tesla’s average price closer to $45,000 for base models
• The cost to produce a solar module in China is $0.15 per watt, vs. $0.40 per watt in the US

For this month’s FTFL, we attempt to measure a “Return on Innovation” in different geographies. We focused on the top 10 tech companies by market cap in the US, China and Europe. To compare across varying business models and accounting methods, we came up with our own metric: 

Return on Innovation = EBITDA / (R&D Expense + Capex)

We looked at the 10-year aggregates across these companies and geographies.

We were floored at how the US and China aggregates are the same and Europe is not far off, even at an order of magnitude difference in dollar amounts. Aggregate R&D + capex spend for the US top 10 in 2024 was 15x over China and 13x over Europe. 

Gotta be a wild coincidence, right? Nope.

This dynamic has been remarkably consistent. Over the past 10 years, there simply has not been much variation in Return on Innovation by geography. While it’s unclear to us why this is, we have a few theories:

• Big companies have been borderless – notwithstanding the constantly-updating tariff news, these massive companies are spending and earning all across the globe, so no single locality will have a major effect.
• Investor return demand is the same everywhere. Perhaps companies tend to gravitate toward similar profitability metrics to attract investors. Taken in the aggregate, these metrics across geographies start to converge.
• Perhaps government policies and regulations can have specific or localized effects, but at scale (with apologies to the great Dr. Ian Malcolm), innovation finds a way.

One thing we noted in the first table – the individual company numbers vary wildly. We traced a few of them over the last 30 years.

It’s like the History of Big Tech in one handy chart.

• Microsoft as the “adult in the room” capitalizing on the Dotcom Bubble
• Apple’s 2008 iPhone release completely transforming the company
• The AI Revolution has NVIDIA looking to surpass the heights of the iPhone

Let’s dig into the AI numbers a bit more. 

In Nvidia’s 2024 10-K, they noted that 34% of their revenue came from 3 unnamed customers. It’s widely believed that these customers are 3 of Alphabet, Amazon, Meta and Microsoft, and that the 4th just missed the 10% threshold to be called out in the filing.

So let’s assume that the 4 companies accounted for 43% of NVIDIA’s revenue, or $56B of $130B. That $56B represents a whopping 25% of Alphabet, Amazon, Meta and Microsoft’s aggregate capex. A huge bet on AI, and that doesn’t even count the billions of direct investment dollars Big Tech has poured into AI companies like OpenAI, Anthropic, Databricks and Scale AI. Nor does it count Apple’s reported $1B purchase of NVIDIA servers.

Back to the chart above, Wall Street estimates that NVIDIA’s 2025 Return on Innovation will skyrocket to 790%. Meanwhile, our other 5 Hyperscalers, whose aggregate Return on Innovation over the last 5 years was 145%, are projected to decline slightly to 133% in 2025. It appears that Wall Street may already be starting to question if Big Tech is overinvesting in AI.

This article appeared in our April 2025 issue of From the Front Lines, Bowen’s roundup of news and trends that educate, inspire and entertain us. Click here to subscribe.