Short answer. Both. The technology is real and durable. Large language models solve genuine problems that owners can use today. The valuations of some AI companies and the intensity of the hype cycle contain a classic bubble that will correct. Whether or not NVIDIA is overvalued has nothing to do with whether AI helps your SMB. Productivity gains are real regardless of stock prices.
The Real Part
LLMs work. They draft, summarize, categorize, and research at machine speed. This is not a demo. It is a durable capability that businesses use every day.
The productivity gain for information workers is measurable. 5 to 15 hours per employee per week is real. That does not evaporate when stock prices correct.
The underlying technology (transformers, large models, RLHF) has been in development for a decade. It is not new. It is now good enough to matter.
The tools will keep improving. Consumer access will keep expanding. This is a permanent shift in how information work happens, similar to how the internet changed communication.
The Bubble Part
Some AI companies are trading at valuations that assume unrealistic growth trajectories. When those assumptions do not materialize, the stocks correct.
The venture capital cycle is producing many AI startups that will not survive. Historically, in every technology wave (dot-com, mobile, crypto, cloud), 80 to 90 percent of startups fail. AI will be similar.
The hype cycle is producing wildly overpromised timelines. AGI in 18 months. AI replacing all knowledge work by 2027. AI curing diseases. Some of these will happen slower than the hype suggests. Some will not happen at all.
None of this means AI is not real. It means the market is pricing some assumptions that will not fully materialize. Bubbles correct. Underlying technology stays.
What This Means For Your SMB
Adopt AI for the productivity gains available today. Do not wait for AGI or for the bubble to burst. Both are wrong reasons to defer.
Do not invest your business in AI startups you do not need. Most of them will not exist in 3 years. The tools you use for work matter. The company you buy stock in is a separate question.
Standardize on tools with financial staying power. Microsoft (Copilot), Google (Gemini), OpenAI, Anthropic, and a handful of others will still be here in 5 years. Smaller AI startups may not.
The technology is a productivity tool for you. Not an investment opportunity. Do not confuse the two.
Frequently Asked Questions
Will AI companies crash? +
Some will. NVIDIA, OpenAI, Anthropic, Google, and Microsoft will likely be here in 5 years. Many smaller AI startups will not. The AI infrastructure spending has some correction risk. The underlying technology does not evaporate.
Should I put my business through AI now or wait? +
Adopt now for the productivity gains. Waiting for 'the perfect AI' is like waiting for the perfect smartphone in 2007 - the technology will keep improving, but you miss compound value by waiting.
Is this like the dot-com bubble? +
Similar in some ways. Real technology, real productivity gains, some overvaluation, many startups that will not survive. The dot-com survivors (Amazon, Google) built enormous durable value. AI will produce similar survivors and similar casualties.
How do I invest in AI safely? +
Not investment advice, but broad index funds catch AI upside without single-company risk. If you must pick individual stocks, established platforms (MSFT, GOOG, META, NVDA at reasonable valuations) beat AI-startup speculation. Consult a licensed financial advisor for actual investment decisions.
