- 41%
- Accuracy
- 37 of 41
- Analyst ranking
- 3
- Predictions scored
- 5
- Still to fall due
All categories ▾
Wrong · 0%And sometime early next year, Tirias Research expects Microsoft support for both desktop and mobile PCs using discrete graphics processing units (GPUs).
What happened. Copilot+ PC status still required a 40+ TOPS NPU through 2025. Microsoft only began letting Copilot+ language-model features run on discrete GPUs (Nvidia RTX 30-series and later) in June 2026, and then only experimentally, through a Windows Insider experimental channel with developer mode on. Nothing happened in 2025. The only movement was an experimental GPU path in mid-2026, which did not make discrete-GPU PCs Copilot+ PCs. If Jay treats the 2026 test as the event, the year-call rule would give Half right. Sources: tomshardware.com, geektopia.es, tomshardware.com.
Source: The AI PC Revolution Is Just Beginning (Forbes) · 'next year' rule
Mostly right · 75%Tirias Research believes that the limited forms of GenAI in use today, such as text-to-text, text-to-speech and text-to-image, will rapidly advance to video, games, and even metaverse generation starting within the next 18 to 24 months
What happened. Video generation arrived inside the window: OpenAI released Sora publicly in December 2024 and Google launched Veo 3 with audio at I/O in May 2025. Game/world generation appeared only as research (Google DeepMind's Genie 2, December 2024); nothing like metaverse generation shipped. The core of the call, generative AI moving to video within 18-24 months, came true with public products from OpenAI and Google. Games came only as research demos and metaverse generation did not happen, so it is not fully right. Sources: openai.com, blog.google, deepmind.google.
Source: GenAI Breaks The Data Center Part II: Moving GenAI To The Edge Through On-Device Computing (Forbes) · 24 months from date made
Wrong · 0%When Tirias Research first forecasted global generative AI demand in 2023, we predicted token output would reach an aggressive 20 trillion tokens ... by the end of 2024.
What happened. By Tirias's own account in the same June 2025 column, 'actual usage surged to a staggering 667 trillion tokens' in 2024 (the article also gives 677 trillion later on); a July 2026 Tirias white paper restated 2024 at 990 trillion. The forecast was about 33 times too low. Forecast of 20 trillion tokens against a first published actual of 667 trillion is a miss of over 95%, far outside the 25% band. McGregor himself says the estimate was 'soon overwhelmed'. Sources: forbes.com, forbes.com.
Source: The Unstoppable Growth Of Generative AI (AI Outlook Part 1) (Forbes, 20 Jun 2025)
Caveat. No independent evidence available. Scored on Tirias Research’s own figures.
Still to fall due5 predictions
- Dec 2030
the annual rate of token generation is expected to skyrocket from 677 trillion in 2024 to 2,092 trillion by the end of 2025 and 77,000 trillion (77 quadrillion) by the end of 2030.
Artificial Intelligence · made Jun 2025 · source - Dec 2025
the annual rate of token generation is expected to skyrocket from 677 trillion in 2024 to 2,092 trillion by the end of 2025 and 77,000 trillion (77 quadrillion) by the end of 2030.
Artificial Intelligence · made Jun 2025 · source - Dec 2028
By the end of 2028, the forecast estimates over 6 billion users or about 90% of smartphone market penetration and over 1 quadrillion annual tokens or a 100X increase.
Artificial Intelligence · made May 2023 · source - Dec 2028
Tirias Research forecasts 2028 data center power consumption of close to 4,250 megawatts, a 212X increase over 2023
Infrastructure & Telecoms · made May 2023 · source - Dec 2028
Tirias Research forecasts that on the current course, generative AI data center server infrastructure plus operating costs will exceed $76 billion by 2028
Infrastructure & Telecoms · made May 2023 · source