Talk Around Town: A 2016 Tech Prophet's AI Report Card Is a Mixed Bag, According to One Analyst

What Folks Are Chattering About

Well, butter my biscuit and call me a prophet — somebody dug up Kevin Kelly's 2016 book 'The Inevitable' and held it up to the 2026 sun to see what shines through, and an independent blogger at taylor.town published what amounts to a one-person report card on the whole enterprise.

The taylor.town analysis argues, in that author's own interpretive framing, that Kelly landed some solid hits, whiffed on others, and got blindsided by a piece of math he couldn't have packed in his truck before it existed — a neat trick even the best forecasters can't pull off.

The Hacker News crowd, as summarized by daily.dev, mostly responded by talking about Ray Kurzweil instead, which is a little like asking someone to judge a cornbread cook-off and watching them wander off to discuss sourdough, but it does reflect that community sentiment was predominantly mixed or skeptical, per daily.dev's own tags.

What Is Actually Known — The Hard Facts

Whatever one thinks of prediction-grading as a sport, some of the underlying numbers are not in dispute: Alphabet's own SEC filings confirm that Google Search revenue came in at $60.4 billion in Q1 2026 and $63.3 billion in Q2 2026, representing year-over-year growth of 19% and 17% respectively, making it unambiguously Alphabet's largest revenue segment.

Alphabet's Q2 2026 earnings materials quote CEO Sundar Pichai describing AI investments as redefining what is possible across the business, though the filings themselves treat AI as a driver of search growth rather than a replacement for it — a distinction that turns out to matter quite a bit when you go to grade the Kelly forecast.

On the cloud-oligopoly side, independent financial analysis from ValueAdd VC and others documents that Microsoft Azure, Google Cloud, and Amazon Web Services do in fact dominate AI infrastructure spending, and together with Meta and Amazon, the four major hyperscalers are projected to spend roughly $325 to $340 billion on AI infrastructure in 2026, according to that analysis.

CNBC and VentureBeat both reported in late April 2026 that OpenAI ended its cloud exclusivity with Microsoft, with its models moving onto Amazon Web Services via the Bedrock platform and potentially expanding to Google Cloud — a real, documented structural shift in how frontier AI models are distributed across the market.

Kelly's Google Forecast: Hit, Miss, or Both at Once?

Here is where the mud gets real thick on the boots: the taylor.town analysis notes that Kelly's 2016 book quoted what he described as a conversation with Larry Page, in which Page allegedly told him 'Oh, we're really making an AI,' and Kelly subsequently predicted that Google's main product by 2026 would be AI rather than search.

The trouble, as this publication reads it, is that calling that prediction either a clean hit or a clean miss is like trying to nail Jell-O to a pine tree — Google Search is still the biggest cash cow in the barn by a country mile, but Alphabet's own filings frame AI as the engine pulling that cow along, which makes both 'correct' and 'incorrect' defensible readings depending on which end of the beast you're measuring.

Forrester's analysis, published in April 2026, put it squarely: generative AI is rebuilding search, and Google is still winning — a framing that is simultaneously a point in Kelly's favor and a point against him, depending on whether you care more about the product label or the revenue line.

The taylor.town author scores this as a partial credit at best, and that framing seems reasonable as editorial analysis, though readers should keep firmly in mind that it is one blogger's call, not a settled verdict — Kelly himself has not, as far as this publication can find, publicly graded his own 2016 call on this specific question.

The Cloud Oligopoly Call: One He Might Brag About at the Cookout

The taylor.town analysis contends that Kelly's prediction of what it describes as an oligarchy of two or three large, general-purpose cloud-based commercial intelligences has a plausible claim to being the strongest call in the bunch, and the independently documented hyperscaler spending figures lend that reading some genuine weight.

OpenAI's departure from Microsoft exclusivity, as reported by CNBC and VentureBeat in April 2026, is consistent with the 'AI like a utility' metaphor Kelly reportedly used in the book — the idea being that AI would flow like electricity through existing cloud infrastructure rather than sitting in one company's barn.

That said, the competitive dynamics are considerably more tangled than a clean oligopoly: OpenAI, Anthropic, and a growing roster of open-weight models are all jostling for position within that infrastructure, which is less a tidy three-way footrace and more like a county fair hog scramble with extra contestants nobody invited.

The 'Narrow AI' Forecast: Trampled by a Transformer

According to the taylor.town analysis, the roughest terrain for Kelly's forecast is his reported claim that the bulk of AI would remain narrow and special-purpose — able to translate languages, for example, but unable to do much else — a prediction the analysis argues was effectively upended by the transformer architecture introduced in 2017, the very year after the book was published.

The taylor.town author frames this as Kelly running smack into what AI researchers call 'the bitter lesson' — the empirical finding that scaling general methods on lots of data tends to beat hand-crafted narrow systems — and it is hard to argue with the observation that large language models capable of broad, open-ended conversation were simply not on the map Kelly was drawing from in 2016.

To be fair to Kelly, the taylor.town analysis itself acknowledges that most deployed AI models are still task-specific by count, so one could defend a weaker version of his prediction — but by public mindshare and investment dollars, the dominant paradigm is now general-purpose LLMs, and that is a meaningful gap between what was predicted and what arrived.

Kelly's Own 2026 Posture: Considerably More Humble

Whatever Kelly's 2016 book projected with confidence, his own Substack post from April 2026 describes the current AI moment as what he calls a sustained, extended period of uncertainty lasting a decade or more — a notably more cautious tune than the inevitabilism his book reportedly sang a decade ago.

That shift in tone, according to Kelly's own Substack writing, does not constitute a self-graded report card on his 2016 forecasts, and this publication is careful not to treat it as one — but it does suggest that even the man who wrote the book is holding his predictions a bit more loosely these days than the title might imply.

Our Analysis: Report Cards Are Tricky When the Teacher Picks the Rubric

Speaking strictly as analysis and not as reported fact: the most durable insight lurking inside this whole grading exercise is the Hacker News community's meta-objection, summarized by daily.dev, that long-horizon tech predictions are often too vague to score fairly — a critique that applies to Kelly's forecasts and to every other entry in the genre simultaneously.

The cloud-oligopoly prediction stands out as the most empirically defensible call precisely because it is structural and measurable: three platforms dominate, the infrastructure spending is documented, and the utility metaphor holds up reasonably well even if the competitive texture is messier than a simple oligopoly.

The Google-search-to-AI call, on the other hand, is a reminder that predictions framed around what a company's 'main product' will be tend to slide around under scrutiny like a greased pig at a county fair — definitional enough to resist falsification but interesting enough to keep people arguing, which may be exactly why this kind of forecasting remains a beloved spectator sport regardless of its predictive track record.