I recently finished listening to the Acquired podcast. Not just the latest episode, but all of them. When I started, I had decided to listen in reverse chronological order, winding my way back through the last 10+ years of episodes.

Walking backward through tech history meant listening to predictions from progressively further back with the absolute benefit of hindsight.

The show’s 2016 year-in-review and 2017 predictions episodes are particularly fascinating time capsules. Looking back now, a decade later, the tech worldview of 2016 wasn’t naive. Many of today’s major themes were already visible. Where the industry stumbled wasn’t on the what, but on the when, the how, and the sequencing.

Here is how some of the predictions, frameworks, and narratives of that era have aged.

An abridged version of this is available in the following X/Twitter thread:

1. The Shifting AI Paradigm: From Blue-Collar to White-Collar

The conversation around AI and labor displacement was just as loud in 2016 as it is today, but the target of concern was entirely different.

  • The 2016 Fear: Automated driving, trucking, logistics, and physical-world automation. The prevailing belief was that blue-collar and transport jobs would be the first to disappear.
  • The Reality: While autonomous trucks and commercial robotaxis are finally starting to roll out, the rollout has been slow, hyper-regional, and heavily gated by regulations. Instead, the most acute AI anxiety today is around software, writing, legal, finance, and white-collar work.

Furthermore, the machine learning discussion back then was thoroughly pre-foundation-model. ML was treated as a domain-specific capability: better ad routing, fraud detection, or voice recognition. What was not central to the conversation was a general-purpose AI product—a simple text box used by hundreds of millions of people for an infinite variety of tasks.

The Death of the Simple “Data Moat” In 2016, the standard thesis was that algorithms would commoditize, and the only true moat would be proprietary data. While data remains crucial, the LLM era proved that moats are far more complex. Today, distribution, compute capital, talent, inference costs, product surface area, and enterprise trust matter just as much as the underlying dataset.

2. The Timeline Trap: Autonomy and XR

If you want to see where tech optimism outpaced reality, look no further than autonomous vehicles and spatial computing.

Autonomous Vehicles

In 2016, many projected that autonomy would completely reshape logistics and rideshare within a few years. A decade later, we are still in the very early innings. The rollout has been safe but slow, limited by geography, operations, and regulatory caution (with continued disengagements that rely on mobile operators). There is a very long and persistent tail to autonomy. At this point, few people doubt that there will be an engineering victory; the failure was in predicting the timeline.

VR / AR

A decade ago, tech circles genuinely believed we were one or two years away from knowing if VR would go mainstream (training, logistics, personal IMAX, gaming, etc). Since then, Meta spent aggressively, the hardware improved substantially, and after a long delay Apple entered the arena. Yet, the dream of XR as a mainstream computing platform remains unsettled. The technology is good enough to satisfy many use cases, but the universal consumer habit didn’t follow at the expected pace.

3. M&A: Company Outcomes Are Not Product Outcomes

Listening to old M&A reactions highlights the massive delta between immediate market reactions and long-term strategic execution.

  • Twitter / X: In 2016, there was dead-serious analysis about Disney acquiring Twitter to pair live sports and media with real-time conversations. Instead, Twitter remained independent, became a global political battleground, was taken private by Elon Musk, rebranded to X, and became deeply integrated into an AI play. It is hard to find a company that had a stranger, more volatile decade.
  • Microsoft / LinkedIn: This was a massive, somewhat polarizing acquisition at the time. In retrospect, it aged beautifully. It integrated seamlessly into Microsoft’s enterprise identity, recruiting, and sales tools, and now serves as a powerful distribution node for AI.
  • Walmart / Jet.com: As a standalone consumer brand, Jet.com dissolved. But as a strategic move, it looks much better than the product outcome alone suggests. It injected e-commerce DNA, talent, and existential urgency into Walmart at a time when competing with Amazon was a matter of survival.

4. Platform Dynamics: Aggregation and the Streaming Wars

In 2016, Ben Thompson’s Aggregation Theory was the dominant lens for understanding the internet. Listening to these episodes, you can hear tech strategy shifting in real-time as people used the framework to understand how Amazon, Facebook, Google, and Netflix were capturing value by owning the consumer relationship and pulling fragmented supply into their orbit. This framework has aged extremely well.

The applications of it, however, evolved in ways few predicted:

  • The Streaming Wars: The battle between Amazon Video and Netflix was originally framed as a pure business model fight (standalone subscription vs. bundled “free” with Prime). Over the decade, that clean battle dissolved into a complex mess of ad-supported tiers, bundles, sports rights management, and high churn—essentially reinventing the mechanics of cable television.
  • Facebook and Media: 2016 was near the peak of Facebook’s influence over digital publishing (Instant Articles, AMP, the pivot to live video). That entire era now feels ancient. While platforms and media still clash, the dream of a Facebook-native publishing ecosystem mostly collapsed, leaving publishers to scramble back toward direct subscriptions and owned distribution.

5. Market Structures and Macro Shifts

The macro-environment and cultural tone of tech media have shifted dramatically over the last ten years.

  • Staying Private Longer: Following Snap’s IPO, there was hope that the era of companies staying private indefinitely was ending. The opposite happened. The private-market machine grew vastly more powerful with mega-rounds, tender offers, and sovereign wealth capital. The public market continues to miss out on a massive portion of early-stage growth. If current trends continue, the next few years are actually primed for several trillion dollar IPOs.
  • The Realization of Physical Fragility: One underrated, highly accurate prediction from the 2016 era was the vulnerability of global supply chains and the impending return of tariffs. The decade that followed was defined by exactly that: supply shocks, reshoring debates, inflation, and a deeply political focus on where hardware is actually manufactured.
  • Cultural and Political Realignment: The political tone of tech media was markedly different post-2016. Major tech events or IPO episodes often opened with editorial statements on politics. Today, tech operators and podcasters often engage in political discourse in more nuanced ways. The broader approach of the industry has largely shifted away from institutional corporate statements toward individual, nuanced, and often less overtly partisan commentary.

The Lesson in Forecasting

Looking back at 2016, the tech ecosystem did not suffer from a lack of imagination. Many of the trends we are living through today were already visible on the horizon.

The breakdown happened in timing, sequencing, and interface.

Knowing that a technology will exist is the easier part. Knowing when the unit economics will make sense, which regulatory hurdles will stall it, what interface will make it intuitive, and which company will capture the value is the harder problem.

That is what makes old predictions useful. They show where the original model of the future was too clean.

I would love to see Ben Gilbert and David Rosenthal do the companion piece to 10 Years of Acquired: not a retrospective on the show (which they have already published), but a retrospective on the forecasts.

Next, I would love to see the harder version: what should we be predicting now?

Which parts of 2036 already seem obvious from here, and which parts are we misreading because we are too focused on the current interface, the current company, the current business model, or the current rate of adoption? That is the version I would be eagar to reread ten years from now.