Apple and Google have announced a “multi-year collaboration” to unlock “innovative new experiences” for Apple users. Read between the lines and the message is clear: Apple is choosing to stay in its own lane, while Google has moved one step closer to winning the AI race.
AI race first battle: model superiority
Late last year, Sam Altman told OpenAI employees to improve ChatGPT in a “code red” memo. The warning came after Google unveiled Gemini 3, its latest-generation AI model.
With Gemini 3 Pro, Google has moved from playing catch-up to genuine front-runner. It appears to have overtaken ChatGPT and Claude across many benchmarks.
While ChatGPT still commands the largest market share, Gemini and Claude are expanding at a much faster pace, suggesting the model gap is closing rapidly.
The real battle: monetarisation and cost control
In 2026, the battle between Google, Anthropic and OpenAI will be beyond who can build a better model, but about distribution, monetisation and being cost efficient.
Google’s Game Plan:
#1 Dominate distribution
Google’s strategy is to make Gemini far more than a standalone chatbot. Through the Apple-Google partnership, Gemini is being woven into the two largest consumer and technology ecosystems in the world, reaching billions of users through products they already use every day.
- Google: Search, YouTube, Android, Google Cloud, etc.
- Apple: iPhone.
#2 Win the cost war
While OpenAI is reportedly planning a US$1 trillion IPO, and Anthropic’s Claude remains a strong technical contender, neither controls the infrastructure economics that now define long-term advantage in AI.
Few investors appreciate that Google’s Tensor Processing Units (TPUs) began development as early as 2015. That decade-long head start is decisive. Based on Semi Analysis, Google’s TPUs are believed to be:
- ~30% cheaper than Nvidia GPUs
- Delivering 2–4x better performance per dollar in comparable workloads
This cost advantage means cheaper model training and higher margins.
OpenAI and Anthropic, both dependent on Nvidia’s supply chain, cannot match Google’s cost base.
#3 Vertical integration
Google is the only player that owns the entire AI stack:
- The models (Gemini)
- The training and inference hardware (TPUs)
- Global data centres
- The enterprise cloud platform
- The consumer distribution channels
This vertical integration allows Google to deploy AI more efficiently than OpenAI or Anthropic. It also gives Google Cloud a compelling value proposition: predictable pricing, TPU-native performance, and AI deeply embedded into enterprise workflows.
OpenAI: trillion-dollar burden
OpenAI’s IPO, potentially one of the largest in history, is already facing an enormous operating cost. HSBC projects that OpenAI’s total data centre rental obligations may approach US$1.4 trillion through 2033.
Against that backdrop, OpenAI’s own forecast of US$100 billion in revenue by 2028 covers a small clip of its annual infrastructure bill. Google won’t have the same level of capex burden.
Anthropic: Strong model, structural limits
Anthropic has positioned itself as the more cautious, safety-focused alternative, and Claude is highly regarded for accuracy and guardrail design. However, it faces two significant disadvantages:
- It relies on other cloud providers, including Google.
- No direct distribution to consumer.
2026: The year Google takes the lead
We expect Google to take the lead in AI in 2026, driven by three forces:
- Faster to commercialise: Google deploys AI across billions of users with trillions of touchpoints that its competitors cannot replicate.
- Model convergence: Gemini is now a very competitive model.
- Hardware economics: Cost advantage on infrastructure give Google an unmatched margin advantage.
How Investors can position
For investors who want exposure to the leaders that are most likely to commercialise AI successfully, ETFs such as HUGE (Magnificent 7+) and WWWW (US Technology) provide meaningful exposure to Alphabet (Google) while capturing the broader ecosystem powering the AI boom. They offer a diversified approach for investors seeking to participate in AI’s next phase.
Disclaimer:
The issuer of units in ETFS Magnificent 7+ ETF (HUGE) (ARSN: 685 356 183) and ETFS US Technology ETF (WWWW) (ARSN: 685 355 971) is the responsible entity of the Fund, being ETF Shares Management Limited (ABN 77 680 639 963, AFSL: 562 766). The product disclosure statement (PDS) and Target Market Determination (TMD) for the Funds contain all of the details of the offer of units in the Fund. Copies of the PDS and TMD are available from ETF Shares Management Limited or at www.etfshares.com.au.
The information provided in this document is general in nature only and does not take into account your personal objectives, financial situation or needs. Before acting on any information in this email, you should consider the appropriateness of the of the information having regards to your objectives, financial situation or needs and consider seeking independent financial, legal, tax and other relevant advice.
Investment in any product issued by ETFS are subject to investment risk, including possible delays in repayment and loss of income and principal invested. The value or return of an investment will fluctuate and an investor may lose some or all of their investment. Past performance is not a reliable indicator of future performance
