Alphabet, the parent company of Google, reported a profit of $34.5 billion for the most recent quarter, outperforming Wall Street expectations, with revenue soaring 48% driven by cloud computing. Alphabet is planning $175 to $185 billion in 2026 capex—far above expectations—to build AI computing capacity. Google Cloud is now running at about $71 billion annually, and that growth is fueling an infrastructure arms race.
Google’s Gemini app has achieved a significant milestone, surpassing 750 million monthly active users. According to data from industry analysts, this rapid adoption places Google’s Gemini among the top applications in terms of active users.
OpenAI’s market share for ChatGPT has declined significantly, falling from 69.1% to 45.3% among daily U.S. mobile app users between January 2025 and January 2026, as competition in the artificial intelligence chatbot sector intensifies. During the same period, Google’s Gemini saw its share increase from 14.7% to 25.1%, while Grok rose from 1.6% to 15.2%. According to data from Apptopia, the overall chatbot market has surged by 152% since last January. Furthermore, user engagement statistics reveal that 20% of chatbot users were utilizing at least two different apps by the end of 2025, compared to just 5% at the end of 2023.
Why do we care?
Google just told Wall Street they’re spending $180 billion on infrastructure this year—60% more than anyone expected. That’s not confidence. That’s an arms race. And arms races have a specific economic logic: you spend because you must, not because the ROI is clear.
Now connect that to the Apptopia data buried in this story. Three years ago, 5% of chatbot users touched multiple apps. Today it’s 20%. The AI interface layer is commoditizing in real-time. ChatGPT lost 24 points of market share while the total market grew 152%—which means OpenAI probably gained users while losing dominance.
This is the pattern that should concern you: the hyperscalers are spending unprecedented capital to build AI infrastructure, while the actual user-facing products built on that infrastructure are becoming interchangeable. The value is accruing to the picks-and-shovels layer, not the application layer.
Here’s the leverage point for MSPs: procurement and liability. As tools commoditize, clients will pay for governance—model selection policy, data handling, audit trails, and security controls. Productize it: an ‘AI Operations & Governance’ managed service with measurable risk and productivity reporting. Include: acceptable-use policy, model/vendor risk reviews, data classification, DLP and logging, prompt/agent change control, and quarterly ROI + risk reporting.
If you’re an MSP building your AI practice around a single vendor—and let’s be honest, for most of you that’s Microsoft because of existing Entra and 365 relationships—you’re accepting platform concentration risk at exactly the moment when your clients will start demanding model flexibility.
The bad decision this story enables: assuming Google’s user growth or Microsoft’s enterprise position means you should double down on one stack. The correct read is that AI tooling is becoming infrastructure. And infrastructure, as every MSP knows, requires diversification and optionality.
The $180 billion tells you the buildout is real. The fragmentation data tells you the moat isn’t. Build accordingly.