Value in Technology Value Chains Tends to Migrate to the App Layer

Slow revenue growth and lower average revenue per account are hardly new concerns for suppliers of consumer access services (mobile or fixed). 

But we should not be surprised, either. 

The rule in technology industries is that economic value tends to migrate upward in the technology stack. Network effects are one reason. But opportunities for customer relationships, loyalty and multiple revenue models also make a big difference. 

Asset	Access provider	Application
Customer relationship	Weak	Strong
User data	Limited	Extensive
Workflow integration	None	Deep
Brand loyalty	Moderate	High
Network effects	Small	Often enormous
Pricing flexibility	Low	High

So in the internet value chain, roughly half of ecosystem revenues accrue to app providers, while access providers (internet service providers, mobile service providers) get between 15 percent and 20 percent. 

Value chain layer	Typical participants	Approx. share of ecosystem revenues	Economic characteristics
User applications & digital services	Google, Meta, Microsoft, Netflix, Salesforce	45–55%	Highest margins and strongest network effects
Commerce & digital platforms	Amazon, Uber	20–25%	Transaction-based economics
Cloud & enabling services	Amazon Web Services, Microsoft Azure, Google Cloud, CDNs	10–15%	Infrastructure with higher value-added
Internet access	ISPs, cable, mobile operators	15–20%	Capital intensive, regulated, slower growth
Passive infrastructure	Towers, fiber REITs, colocation	5–10%	Stable but utility-like returns

The economic principle is simple:

• Infrastructure competes on capacity

• applications compete on customer outcomes.

Capacity usually becomes abundant, and abundance reduces pricing power.  Solutions for customer problems remain “scarce,” in the sense that customers gravitate to a relatively few apps and tend to stick with them over time. 

And scarcity supports pricing power. 

Economic force	Internet example	AI analogy
Infrastructure becomes commoditized	Broadband, fiber and mobile access become widely available	GPU clusters eventually become standardized compute utilities
User attention concentrates	Search, social media, streaming dominate consumer engagement	AI assistants and vertical AI agents become primary interfaces
Switching costs increase higher in stack	Users stay with Gmail, Office 365, Salesforce—not because of ISP	Users remain with AI workflow platforms because of memory, integrations and data
Network effects strongest near users	Facebook, YouTube, Amazon Marketplace	OpenAI ecosystem, enterprise agent platforms, developer ecosystems
Pricing power follows differentiation	ISP sells Mbps; applications sell outcomes	GPU provider sells tokens; applications sell productivity or decisions
Marginal cost falls faster below than above	Network capacity continually gets cheaper	Compute cost falls faster than value of specialized applications

In the AI ecosystem, similar value chain effects should happen. Value should accrue heavily at the app layer. 

AI layer	Future revenue share	Why
AI applications and agents	40–50%	Own workflows and customer relationships
Vertical enterprise software	20–25%	Industry-specific solutions
Foundation model providers	10–20%	Models become more competitive over time
AI cloud infrastructure	10–15%	Compute utility with economies of scale
Hardware (GPUs, networking)	5–10%	Hardware normalizes after supply shortages
Power and facilities	3–8%	Necessary but infrastructure economics

SpaceX IPO Estimates $26.5 Trillion AI Addressable Market

Not the sort of company mission one normally sees in an S-1 (initial public offering) filing!

On the other hand, the expected future revenue leans on artificial intelligence, not space or connectivity. 

So such goals as “colonizing Mars” are essentially the sizzle on the steak. 

SpaceX S1 filing 

SpaceX estimates a total addressable market of $28.5 trillion:

• $370B in "Space" (space-enabled solutions including Starship (fully reusable booster), lunar economy, point-to-point Earth transport, in-orbit manufacturing)

• $1.6T in Connectivity (Starlink Broadband ~$870B + Starlink Mobile ~$740B, plus enterprise/government)

• $26.5T in AI (compute infrastructure (terrestrial + orbital), subscriptions, advertising, enterprise applications, Grok models, consumer/enterprise/government adoption, X platform monetization. 

Year	Total Revenue	Starlink/Connectivity	Launch/Space	AI/xAI
2025 (Actual)	~$18.7B	~$11.4B	~$4.1B	~$3.2B
2026 (Est.)	~$20–28.5B	Strong growth (e.g., $18–20B+)	Stable/moderate growth	Growing but capex-heavy
Longer-term (2030+)	$100B+	Dominant	Scaled via Starship	Major contributor

Neoclouds and CLECs

For some of us who were active in the competitive local exchange carrier market around the time of the passage of The Telecommunications Act of 1996, neocloud providers such as CoreWeave, Nebius and many others seem to present a market opportunity that is temporary, if potentially lucrative in the short term. 

Though price arbitrage was the temporary CLEC opportunity, shortages of high-performance computing (graphics processing units and other accelerators) are the opportunity for neocloud providers.

A perhaps-lucrative but temporary market window seems to exist for neoclouds, as it once did for CLECs. 

By mandating that incumbent local exchange carriers unbundle their network elements and lease them at attractive wholesale rates to competitors at regulated rates, Congress effectively handed competitive local exchange carriers (CLECs) a business model: arbitrage the gap between the regulated wholesale price of network access and the retail price customers would pay.

But the discounts ultimately ended and the access market eventually shifted to broadband access on owned facilities. The wholesale model effectively collapsed for most CLECs. 

The generative artificial intelligence boom created excess demand for GPUs. 

So the neocloud model is structurally arbitrage: 

• As GPU supply is constrained, offer “GPU as a service”

• Sell access to that resource at a margin, reselling compute

• Build customer relationships before the incumbents close the gap.

CoreWeave, for example had a simple price pitch: 

• we have H100s

• we're GPU-native

• we'll get you capacity faster and cheaper than AWS or Azure. 

Dimension	CLECs (1996–2002)	Neoclouds (2022–?)
Enabling condition	Regulatory mandate opening ILEC networks	GPU supply shock creating hyperscaler rationing
Capital model	Debt-heavy buildout of switching infrastructure	Equity/debt-heavy GPU cluster acquisition
Competitive advantage	Access to regulated wholesale inputs	Early access to scarce NVIDIA allocations
Customer value prop	Cheaper/faster local access	Faster GPU availability, simpler pricing
Incumbent response	Network upgrade, litigation, lobbying	Massive capex, custom silicon, long-term NVIDIA contracts
Structural vulnerability	Unbundling obligations could be reversed	GPU scarcity is inherently temporary
Timeline pressure	~5 years before model collapsed	Likely 3–6 years before hyperscalers close gap

Of course, markets eventually will normalize:

• Nvidia has boosted production of H100s and is ramping B200/B300 series

• Hyperscalers have developed  custom silicon (Google's TPU v5, AWS's Trainium 2 and Inferentia, Microsoft's Maia, and Meta's MTIA)

• Hyperscaler capex is going to be hard to beat, long term

• The software stack advantage will benefit AWS, Google Cloud and Azure

• Customer lock-in dynamics favor hyperscalers.

Of course, history likely rhymes rather than repeating.

The neocloud endgame probably looks similar to the CLEC industry in many ways:

• Most will struggle as GPU spot prices normalize and hyperscaler capacity floods the market (2025–2027)

• A few might be acquired

• One or two may find durable niches

• But hyperscalers likely will dominate the enterprise AI compute market by 2028–2030.

The CLEC parallel is perhaps a reminder that cyclical scarcity is not long-term structural advantage. 

The neoclouds that survive will be those that use the current window not just to sell GPUs, but to build something (software, relationships, operational expertise or specialized capability) that persists after the scarcity evaporates. 

That will be hard to do. 

As investors, we might make some money on neocloud providers in the near term. But the CLEC experience might temper enthusiasm for some of us.

Debating Amazon Leo Objectives

Amazon’s objectives with Leo are debated. 

Is this a standalone telecom business or a strategic infrastructure layer feeding higher-margin businesses (likely AWS)?

The possible motives are complicated as Amazon often talks like a “margin hunter,”  but often acts like a scale builder that tolerates thin margins for a time. 

The trick is that Amazon usually tries to separate where value is created from where it is captured. 

Amazon repeatedly enters markets characterized by low margin and high margin, so “margin” is not the primary consideration.

The effort to find “moats” or bottlenecks where value is extracted, and sometimes a low-margin business can lead to a high-margin moat. 

Layer	Characteristic	Amazon Behavior
Customer-facing layer	Huge TAM, fragmented, price-sensitive	Compete aggressively, often low margin
Infrastructure / platform layer	High fixed cost, scalable, defensible	Invest heavily, aim for high margin long term
Data / control layer	Feedback loops, optimization	Build moats that others can’t replicate

The point is that Amazon doesn’t mind entering a low-margin market if it helps it own a high-margin layer underneath or adjacent to it.

Also, “high capital investment” can be a feature, not a bug:

• High CapEx deters competitors

• Once built, marginal costs drop sharply

• Scale converts fixed costs into a profit flywheel

• Infrastructure can support multiple businesses

• Pricing power eventually comes, once dominance is achieved.

So huge capex commitments are consistent with Amazon’s playbook, if Amazon believes it can control a bottleneck layer.

“Is this a high-margin or low-margin business?” might not be the right question for Amazon leaders, who likely are asking:

• Can we own a critical layer?

• Does this scale globally?

• Does it reinforce our existing flywheels?

• Can we improve cost structure vs incumbents?

• Is there a hidden high-margin component?

So the larger picture is often not the immediate or obvious business, but the ability to create leverage elsewhere. Consumer initiatives such as e-commerce; devices or streaming then can be viewed as demand aggregators and ecosystem lock-in creators that drive revenue indirectly (advertising, cross selling, subscriber lock in).

Enterprise infrastructure plays such as AWS or logistics might be better examples of direct, high margin initiatives.

The thing about Leo is where it fits. From one point of view, consumer telecom is a low-margin, highly-competitive business with high regulatory conditions, low innovation and low growth rates. 

So why even consider it?

Amazon probably envisions non-obvious leverage points:

• Edge computing

• Private 5G networks

• Carrier cloud

• Where Amazon captures high-margin compute, not connectivity

• With different value drivers in consumer and business markets.

Owning a connectivity service could:

• Reduce internal costs

• Improve performance (latency, reliability)

• Be bundled with Prime and devices to

• Drive usage of AWS, the advertising platform and e-commerce

• Support IoT connectivity (devices, logistics, smart home). 

Framed that way, Leo might be viewed as a platform layer supporting:

• Edge cloud

• AWS (compute plus connectivity)

• Telcos as customers

• Prime average revenue per user or account

• Customer retention and acquisition

To be sure, execution will matter. But, in theory, Leo is not directly about high margin. It is about control of what is likely to be a low-margin feature of a higher-margin ecosystem. 

Amazon’s explicit framing is straightforward:

• Create a global broadband access business

• Serving “tens of millions of customers” globally

• in “unserved and underserved” markets

• Offers private connectivity directly into AWS

• for enterprise, government, and telecom customers.

So AWS integration, enterprise and government use cases and private networks might be key, not “consumer telecom.”

Leo then is a connectivity extension of AWS. 

But there are clear risks and some skeptics. 

Optimistically, Leo extends AWS to the edge of the network. 

On the other hand, it is a near-term drag on earnings, in a business with tough economics and financial returns that could take some time to develop.

So it might matter hugely whether Leo can generate AWS pull-through; enterprise demand and other ecosystem upsides. 

Also, how long this takes could matter. 

Layer	Role	Margin Potential
Consumer broadband (Leo ISP)	Distribution / scale	Low
Enterprise connectivity	Premium services	Medium
AWS integration layer	Data + compute + control	High
Ecosystem effects (Prime, commerce, ads)	Indirect monetization	Very high

Sure, it’s risky. But some will point to past Amazon initiatives based on entry into low-margin businesses that provided moats:

• Retail → low margin → enabled AWS scale

• Devices → low margin → enabled ecosystem lock-in

• Logistics → low margin → enabled marketplace dominance.

Leo arguably fits the pattern, optimists will argue. It’s about high-margin AWS, not low-margin telecom. Skeptics will worry about the execution risk.