The AI Cloud's Dirty Secret: Is It All Hype and No Substance?
The hidden truth about the AI Cloud is a dirty secret.
If you are wondering if AI is all hype and no substance then Bessemer's State of the Cloud 2024 Report is only fueling that hype.
What's a AI Native Use-case that only AI Cloud can Support? None!
Here is my contrarian view debunking the popular Hype Report of this Year:
1. Legacy Cloud's Resilience
While AI is transforming the landscape, legacy cloud systems will continue to play a crucial role in the business world.
A 2023 survey by Flexera showed that 80% of enterprises have a hybrid cloud strategy. Meaning they use both public and private clouds.
This indicates that legacy systems are not abandoned anytime sooner. But rather integrated with newer AI-powered solutions.
Gartner predicts that by 2025, 85% of organizations will embrace a cloud-first principle. But that doesn't mean a complete abandonment of legacy systems.
Many companies will opt for a phased approach. Migrating workloads to the cloud while maintaining critical legacy systems.
Btw, The data privacy issue was never addressed by the AI folks. Until that's sorted and accepted the legacy cloud is reigning king!
A 2023 study by Cisco found that 60% of organizations are concerned about data privacy in AI applications. Issues like data bias, transparency, and the potential for misuse of AI-generated data raise significant concerns for businesses, particularly those in regulated industries.
Wait! Even if that is not sorted, AI Cloud is nothing but models on Legacy Cloud, isn't it?
LoL, did Bessemer lose it's shit or a bad wordplay?
Even the most advanced AI models, like GPT-4, run on massive data centers powered by legacy infrastructure. Some advanced GPU's run next to them as GPUs alone can't do shit.
The underlying hardware and software that support AI relies on traditional cloud technologies.
2. AI Model Commoditization and Cost Concerns
The prediction that AI models will either become commoditized or dominated by a few giants is crazy.
The AI model landscape is evolving, with new architectures and approaches emerging. First it's transformers, then LLMs, RAG and then DSP and then on top Multi-Modal LLMs and it's only the start.
This suggests that a diverse ecosystem of models is more likely than a few dominant players. Like Meta’s Llama.
Open-source models like Meta's Llama are challenging the dominance of closed-source models.
Making the AI offering a more accessible and cost-effective option for smaller players.
The costs associated with developing, training, and maintaining complex AI models could become sky high for smaller players.
A 2022 report by OpenAI estimated that training a large language model can cost millions of dollars. Smaller companies may struggle to compete with the resources of tech giants like Google and Microsoft.
If SMB's who innovate the most in any space get stuck over cost, it's going to be another Blockchain crash trend with AI. Where there is no adoption, there is no future.
3. Overemphasis on AI Developers
They have underplayed the intent creator over the intent executor. Sounds lame actually to remove the human equation.
While AI tools can automate certain tasks, they still rely on human input and guidance. The most successful AI applications are those that augment human capabilities, not replace them.
A study by McKinsey found that the most significant impact of AI will be on jobs that need creativity and problem-solving skills. Mot those that AI can automate overnight.
4. Vertical AI's Uncertain Future
The future if AI is Generalistic than specialized for Verticals. AI will be better in filling the missing context much better over time.
There are successful vertical AI applications, in healthcare and finance, but the future is Generalist AI models.
These models could outperform specialized models in many tasks, challenging the need for vertical AI.
5. Consumer Cloud's Sustainability
Yes, AI-powered consumer cloud applications are gaining traction. But their long-term sustainability remains uncertain.
Many AI-powered consumer apps struggle to monetize and rely on venture capital funding to stay afloat. This raises questions about their long-term viability.
The consumer market is competitive, with new apps and emerging trends. AI-powered apps will need to innovate and adapt to maintain user interest and generate sustainable revenue.
—
Regardless, I’m sure Bessemer will start writing an Anti-State of Cloud reports soon like it did with its Anti-Portfolio which is much more sensible.
P.S. Did I write it with the help of AI? Yes, some of it but I’m still here, not replaced by an AI and trust me this blog is on legacy cloud infrastructure.