Microsoft's Diverging Path: Rethinking AI Collaboration Beyond OpenAI

Ah, the tech bromance of the decade: Microsoft and OpenAI. With a $13 billion investment in OpenAI, Microsoft has been one of the most ardent proponents of AI innovation. From GPT-powered Bing to GitHub Copilot to the enterprise-focused Microsoft 365 Copilot, their collaboration has reshaped the productivity landscape. But as we cross over into 2024, reports suggest clouds are forming over this once enviable partnership. Microsoft, it seems, is considering integrating third-party AI models into its beloved Copilot 365—and potentially doing so independently of OpenAI. Why? Because GPT-4, for all its brilliance, is apparently too slow and too expensive.
Let's unpack what this might mean—not just for Microsoft and OpenAI, but for the users who heavily rely on Copilot for an AI-enhanced workflow.

---

What's the Buzz Around Microsoft Moving Beyond OpenAI?​

Let’s get straight to the meat of this report. The chatter began when Reuters released a story claiming that Microsoft is on the hunt for alternative AI collaborators. The driving factor? OpenAI's GPT-4 isn't cutting it for Copilot 365 users. Sure, the Generative Pre-trained Transformer 4.0 is all glitz and glam when it comes to conversational intelligence, but when tasked with enterprise-grade productivity—an arena where speed and scalability rule—cracks seem to appear.
Concerns include:

• Exorbitant Costs: At an unprecedented scale like that of Microsoft 365 Copilot, using OpenAI's GPT-4 racks up significant operational expenses. With features like workflow automation, meeting summaries, and email drafting, Microsoft has to foot a massive bill for every word Copilot generates. For OpenAI, maintaining the computational power for GPT-4 isn't cheap either, leading both companies to an uncomfortable financial tightrope.
• Performance Woes: Some clients, as reported, found the AI tool not functioning as expected nearly 75% of the time. Oof. Imagine paying $30 per user per month for something that may or may not work. Complaints are mounting: users say productivity is not soaring as promised, and scalability issues exacerbate the problem.

---

Inside the Strained "Tech Bromance" with OpenAI​

Remember those super-tight partnerships college roommates share, pledging to stick together no matter what? That’s Microsoft and OpenAI… or at least it used to be. Microsoft gained early access to OpenAI's AI models (including the much-hyped generative AI GPT-series), and in exchange, threw heaps of money at the AI lab to accelerate innovation. The partnership fueled products like Microsoft 365 Copilot, which integrates AI into PowerPoint, Word, Teams, and more.
But friendships can be complicated, especially when money—and cosmic ambitions like achieving "Artificial General Intelligence" (AGI)—get involved. AGI, for those unfamiliar, refers to AI systems so advanced that they can perform any intellectual task a human being can with equal or greater proficiency. If OpenAI achieves this milestone, their agreement with Microsoft could dissolve. OpenAI is rumored to be making quicker progress toward AGI than anyone anticipated.
Microsoft doesn't want to be left stranded in that scenario, especially if OpenAI becomes a runaway success. Moreover, OpenAI's hefty computational requirements are becoming a point of friction, potentially slowing down the AGI dream.

---

Why Third-Party AI Models Make Sense for Microsoft​

Adding third-party AI models to Copilot sounds like a smart hedge strategy. Think of this as Microsoft saying, "Let's not put all our eggs in OpenAI's basket." Here's why this approach is worth exploring:

1. Cost Optimization​

Outsourcing to multiple AI providers—or even developing proprietary models—would give Microsoft cost flexibility. OpenAI’s GPT-4 is undeniably top-tier, but its high implementation price makes it less attractive for enterprise-wide scalability. Microsoft has to keep its customers happy by avoiding price hikes while still delivering top-notch services.

2. Performance Issues​

When Copilot doesn't work as expected, it's not a minor gripe—it can disrupt workflows for teams and diminish trust in Microsoft's AI capabilities. By diversifying AI providers, Microsoft can choose the best model suited for specific tasks. For instance:

• Text summarization may be handled by one provider.
• Data visualization tasks in Excel could lean on another.
• Conversational AI responses, OpenAI or otherwise, might still stay in play.

3. Future-Proofing Against AGI Uncertainty​

If OpenAI does hit AGI and decides to either pivot from its partnership with Microsoft or comes to dominate the AI field in unprecedented ways, Microsoft will have a fallback. Playing it safe with multiple partners ensures they won’t lose all their competitive edge in a post-AGI world.

---

The Struggles of Copilot 365 So Far​

Microsoft Copilot 365, when announced, sounded like productivity magic. It promised to do everything except fetch your morning coffee. Features like summarizing emails, analyzing trends in your data, and automating repetitive tasks theoretically made it indispensable for the modern office. However, the product rollout hasn’t been smooth sailing:

• "Gimmick" Accusations: Internally, some Microsoft executives reportedly called these tools “gimmicky,” suggesting the solutions lack the polish Microsoft frequently touts.
• End-User Frustrations: If Copilot appears clunky three times out of four, users are less likely to come back. Even businesses willing to stomach high sticker prices are demanding consistency—and rightly so.

These aren't isolated tech hiccups. Poor adoption translates to dissatisfied enterprise customers and potentially hinders Microsoft's first-mover advantage in AI-enabled productivity tools.

---

What Happens Next?​

There's a lot to unpack in this evolving story. Here’s what could be on the horizon:

1. Expanded AI Collaborations​

By involving other third-party players or doubling down on its own in-house AI talent, Microsoft can not only weaken its dependency on OpenAI but create a thriving ecosystem of AI solutions. Expect to see announcements where Satya Nadella name-drops AI startups across Europe, Asia, and Silicon Valley.

2. OpenAI’s Next Play​

Banks burning cash eventually face hard decisions. OpenAI reportedly faces financial difficulties, with potential losses of $5 billion within 12 months. If OpenAI reshapes its roadmap to remain afloat, it could spell further strain on the Microsoft partnership. Or, you know, everything solves itself by a deus ex machina. (We’ll keep you posted.)

3. Evolution of Copilot Pricing​

Keep an eye out for potential tweaks to Copilot pricing. Microsoft might roll out tiered systems—think AI lite models, premium integrations, etc.—to cater to businesses balking at $30 per seat.

---

A Watershed Moment for Generative AI in Productivity​

Is Generative AI ready to meet high enterprise expectations? The answer seems complicated. AI tools are seductive in their promise but finicky in execution—a shiny new car with a sputtering engine. If OpenAI and Microsoft part ways, it may be the best move for both of them in the long run.
The situation is also a valuable reminder for users and developers alike: AI isn’t some omniscient magic wand that effortlessly solves all problems. It’s an evolving tech landscape that needs refinement, investment, and correct application.
But hey, the bigger question is… if you’re a Copilot 365 user, what’s been your experience so far? We'd love to hear about it in the forums!

---

Source: Windows Central Microsoft wants to add third-party AI models to Copilot 365 without OpenAI's help