Apple reportedly agreed $1B annual deal with Google for Gemini technology access

3 weeks ago 29

The iPhone maker is leaning harder on Google's custom chips to build its AI models, extending a partnership that started in 2024

Apple Inc. (AAPL) price

Full chart and stats

Loading chart…

Market cap $5.1T (shares as of 2026-07-17)

P/E (trailing) 39.8 (EPS to 2026-06-27)

Annual revenue $416.2B USD (FY ended 2025-09-27) 10-K
Annual revenue growth +6.4% (FY ended 2025-09-27 vs FY ended 2024-09-28) 10-K
Annual net income 112,010,000,000 USD (FY ended 2025-09-27, attributable to the company) 10-K

Annual operating margin 32.0% (FY ended 2025-09-27)

Forward P/E Unavailable: no dated forecast source

Gloria daily close for session Oct 27, 2026. Market cap and trailing P/E use this close with the SEC share total and trailing diluted earnings, each dated next to its figure. Annual revenue, revenue growth, net income and margin show their own fiscal period and link the 10-K. Current context may postdate this article.

Apple Inc. and its peers

Token vs underlying stock

Current comparison for AAPL. Historical observations may postdate this article.

AAPLx token unit vs AAPL share · USD

$348.00Oct 27 UTC

+12.1% over displayed period

● AAPLx token┄ AAPL stock: $347.10

Aug 4, 2026: token $310.34 at 20:00 UTC; stock close $309.38 Aug 5, 2026: token $311.72 at 20:00 UTC; stock close $311.00 Aug 6, 2026: token $312.78 at 20:00 UTC; stock close $312.41 Aug 7, 2026: token $313.83 at 20:00 UTC; stock close $313.33 Aug 10, 2026: token $308.62 at 20:00 UTC; stock close $308.26 Aug 11, 2026: token $305.37 at 20:00 UTC; stock close $304.91 Aug 12, 2026: token $302.83 at 20:00 UTC; stock close $302.25 Aug 13, 2026: token $305.79 at 20:00 UTC; stock close $305.26 Aug 14, 2026: token $306.96 at 20:00 UTC; stock close $305.93 Aug 17, 2026: token $306.25 at 20:00 UTC; stock close $305.59 Aug 18, 2026: token $311.31 at 20:00 UTC; stock close $310.03 Aug 19, 2026: token $317.34 at 20:00 UTC; stock close $316.83 Aug 20, 2026: token $311.74 at 20:00 UTC; stock close $311.30 Aug 21, 2026: token $310.51 at 20:00 UTC; stock close $309.35 Aug 24, 2026: token $311.44 at 20:00 UTC; stock close $310.34 Aug 25, 2026: token $310.87 at 20:00 UTC; stock close $309.90 Aug 26, 2026: token $314.52 at 20:00 UTC; stock close $313.45 Aug 27, 2026: token $315.05 at 20:00 UTC; stock close $314.58 Aug 28, 2026: token $320.36 at 20:00 UTC; stock close $319.70 Aug 31, 2026: token $317.65 at 20:00 UTC; stock close $316.85 Sep 1, 2026: token $325.59 at 20:00 UTC; stock close $325.13 Sep 2, 2026: token $325.80 at 20:00 UTC; stock close $324.96 Sep 3, 2026: token $329.13 at 20:00 UTC; stock close $328.21 Sep 4, 2026: token $321.11 at 20:00 UTC; stock close $319.97 Sep 8, 2026: token $316.85 at 20:00 UTC; stock close $316.22 Sep 9, 2026: token $316.30 at 20:00 UTC; stock close $315.34 Sep 10, 2026: token $327.20 at 20:00 UTC; stock close $326.57 Sep 11, 2026: token $333.36 at 20:00 UTC; stock close $332.27 Sep 14, 2026: token $334.25 at 20:00 UTC; stock close $333.08 Sep 15, 2026: token $331.88 at 20:00 UTC; stock close $331.34 Sep 16, 2026: token $333.36 at 20:00 UTC; stock close $332.41 Sep 17, 2026: token $337.97 at 20:00 UTC; stock close $337.00 Sep 18, 2026: token $337.26 at 20:00 UTC; stock close $336.13 Sep 21, 2026: token $340.11 at 20:00 UTC; stock close $338.98 Sep 22, 2026: token $340.64 at 20:00 UTC; stock close $339.75 Sep 23, 2026: token $337.57 at 20:00 UTC; stock close $337.02 Sep 24, 2026: token $337.27 at 20:00 UTC; stock close $335.92 Sep 25, 2026: token $341.92 at 20:00 UTC; stock close $341.07 Sep 28, 2026: token $339.71 at 20:00 UTC; stock close $338.40 Sep 29, 2026: token $330.45 at 20:00 UTC; stock close $329.40 Sep 30, 2026: token $334.56 at 20:00 UTC; stock close $333.02 Oct 1, 2026: token $331.46 at 20:00 UTC; stock close $330.32 Oct 2, 2026: token $334.68 at 20:00 UTC; stock close $333.69 Oct 5, 2026: token $333.54 at 20:00 UTC; stock close $332.89 Oct 6, 2026: token $334.55 at 20:00 UTC; stock close $333.63 Oct 7, 2026: token $337.28 at 20:00 UTC; stock close $336.67 Oct 8, 2026: token $341.31 at 20:00 UTC; stock close $340.42 Oct 9, 2026: token $337.58 at 20:00 UTC; stock close $336.64 Oct 12, 2026: token $341.95 at 20:00 UTC; stock close $340.97 Oct 13, 2026: token $342.44 at 20:00 UTC; stock close $341.27 Oct 14, 2026: token $344.06 at 20:00 UTC; stock close $342.96 Oct 15, 2026: token $343.24 at 20:00 UTC; stock close $342.00 Oct 16, 2026: token $344.31 at 20:00 UTC; stock close $343.46 Oct 19, 2026: token $344.18 at 20:00 UTC; stock close $342.94 Oct 20, 2026: token $340.65 at 20:00 UTC; stock close $339.50 Oct 21, 2026: token $344.81 at 20:00 UTC; stock close $343.99 Oct 22, 2026: token $349.21 at 20:00 UTC; stock close $348.22 Oct 23, 2026: token $348.46 at 20:00 UTC; stock close $347.73 Oct 26, 2026: token $344.87 at 20:00 UTC; stock close $343.91 Oct 27, 2026: token $348.00 at 20:00 UTC; stock close $347.10 Aug 4$302.25 to $349.21Oct 27

Apple is scaling up pre-training of its foundation models on Google’s Tensor Processing Units. That puts some of the most expensive work in Apple’s AI strategy on chips designed by one of its biggest rivals.

What Apple is actually doing

Pre-training is the heavy-lifting phase of building a large language model. That phase eats compute, which makes the choice of hardware a strategic decision rather than a procurement detail.

Apple’s choice is Google’s TPUs. These are custom chips Google built specifically for machine learning workloads, rather than general-purpose graphics processors adapted for AI.

Apple’s first AFM server model was trained on a cluster of 8,192 TPU v4 chips.

The on-device model, the smaller one designed to run locally on Apple hardware, was built on a 2,048-chip slice of TPU v5p chips.

The performance gains were not trivial. The move gave Apple more than 2x the floating point operations per second, or FLOPS, along with triple the high bandwidth memory.

The chips could do far more math per second and hold far more data close at hand while doing it. The result was training time for large language models cut by up to 2.8 times.

How Apple ended up here

The story starts in July 2024. That’s when Apple revealed it had trained its initial Apple Foundation Models, known as AFM, on Google Cloud TPUs instead of the NVIDIA GPUs most of the industry was using.

Availability was part of the reasoning. The shift came amid NVIDIA H100 supply issues, combined with what the research describes as the superior performance and cost-effectiveness of Google’s TPUs for Apple’s needs.

In January 2026, Apple entered a multiyear agreement with Google for access to Gemini technology, a deal reportedly worth $1 billion annually.

Then, in June 2026, Apple announced its third-generation AFM models. That release marked custom development work done with Google.

Not everything runs on TPUs, though. One of the third-generation models, AFM 3 Cloud Pro, operates on NVIDIA Blackwell GPUs inside Google Cloud.

Apple has also been consistent on one point throughout. The company maintains that no private user data was used to train its foundation models, and it says it honors publisher opt-out preferences.

What this means

The most immediate takeaway is that Google has become a load-bearing pillar of Apple’s AI effort. Between TPU-based training, Google Cloud infrastructure and the Gemini licensing deal, Apple’s AI roadmap now runs through Mountain View in several places at once.

AFM 3 Cloud Pro running on Blackwell GPUs shows Apple is still buying into NVIDIA’s ecosystem, just through Google Cloud rather than as a direct anchor customer for its training runs.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.

Previous storyAlphabet says Google Search queries are growing at their fastest pace in five years
Read Entire Article